1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <crypto/hash.h> 26 27 #include <linux/fscrypt.h> 28 29 #ifdef CONFIG_F2FS_CHECK_FS 30 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 31 #else 32 #define f2fs_bug_on(sbi, condition) \ 33 do { \ 34 if (unlikely(condition)) { \ 35 WARN_ON(1); \ 36 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 37 } \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_MAX, 58 }; 59 60 #ifdef CONFIG_F2FS_FAULT_INJECTION 61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 62 63 struct f2fs_fault_info { 64 atomic_t inject_ops; 65 unsigned int inject_rate; 66 unsigned int inject_type; 67 }; 68 69 extern const char *f2fs_fault_name[FAULT_MAX]; 70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 71 #endif 72 73 /* 74 * For mount options 75 */ 76 #define F2FS_MOUNT_BG_GC 0x00000001 77 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 78 #define F2FS_MOUNT_DISCARD 0x00000004 79 #define F2FS_MOUNT_NOHEAP 0x00000008 80 #define F2FS_MOUNT_XATTR_USER 0x00000010 81 #define F2FS_MOUNT_POSIX_ACL 0x00000020 82 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 83 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 84 #define F2FS_MOUNT_INLINE_DATA 0x00000100 85 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 86 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 87 #define F2FS_MOUNT_NOBARRIER 0x00000800 88 #define F2FS_MOUNT_FASTBOOT 0x00001000 89 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 90 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000 91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 93 #define F2FS_MOUNT_ADAPTIVE 0x00020000 94 #define F2FS_MOUNT_LFS 0x00040000 95 #define F2FS_MOUNT_USRQUOTA 0x00080000 96 #define F2FS_MOUNT_GRPQUOTA 0x00100000 97 #define F2FS_MOUNT_PRJQUOTA 0x00200000 98 #define F2FS_MOUNT_QUOTA 0x00400000 99 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 100 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 101 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 bool test_dummy_encryption; /* test dummy encryption */ 139 block_t unusable_cap; /* Amount of space allowed to be 140 * unusable when disabling checkpoint 141 */ 142 }; 143 144 #define F2FS_FEATURE_ENCRYPT 0x0001 145 #define F2FS_FEATURE_BLKZONED 0x0002 146 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 147 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 148 #define F2FS_FEATURE_PRJQUOTA 0x0010 149 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 150 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 151 #define F2FS_FEATURE_QUOTA_INO 0x0080 152 #define F2FS_FEATURE_INODE_CRTIME 0x0100 153 #define F2FS_FEATURE_LOST_FOUND 0x0200 154 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */ 155 #define F2FS_FEATURE_SB_CHKSUM 0x0800 156 157 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 158 ((raw_super->feature & cpu_to_le32(mask)) != 0) 159 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 160 #define F2FS_SET_FEATURE(sbi, mask) \ 161 (sbi->raw_super->feature |= cpu_to_le32(mask)) 162 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 163 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 164 165 /* 166 * Default values for user and/or group using reserved blocks 167 */ 168 #define F2FS_DEF_RESUID 0 169 #define F2FS_DEF_RESGID 0 170 171 /* 172 * For checkpoint manager 173 */ 174 enum { 175 NAT_BITMAP, 176 SIT_BITMAP 177 }; 178 179 #define CP_UMOUNT 0x00000001 180 #define CP_FASTBOOT 0x00000002 181 #define CP_SYNC 0x00000004 182 #define CP_RECOVERY 0x00000008 183 #define CP_DISCARD 0x00000010 184 #define CP_TRIMMED 0x00000020 185 #define CP_PAUSE 0x00000040 186 187 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 188 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 189 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 190 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 191 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 192 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 193 #define DEF_CP_INTERVAL 60 /* 60 secs */ 194 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 195 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 196 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 197 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 198 199 struct cp_control { 200 int reason; 201 __u64 trim_start; 202 __u64 trim_end; 203 __u64 trim_minlen; 204 }; 205 206 /* 207 * indicate meta/data type 208 */ 209 enum { 210 META_CP, 211 META_NAT, 212 META_SIT, 213 META_SSA, 214 META_MAX, 215 META_POR, 216 DATA_GENERIC, /* check range only */ 217 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 218 DATA_GENERIC_ENHANCE_READ, /* 219 * strong check on range and segment 220 * bitmap but no warning due to race 221 * condition of read on truncated area 222 * by extent_cache 223 */ 224 META_GENERIC, 225 }; 226 227 /* for the list of ino */ 228 enum { 229 ORPHAN_INO, /* for orphan ino list */ 230 APPEND_INO, /* for append ino list */ 231 UPDATE_INO, /* for update ino list */ 232 TRANS_DIR_INO, /* for trasactions dir ino list */ 233 FLUSH_INO, /* for multiple device flushing */ 234 MAX_INO_ENTRY, /* max. list */ 235 }; 236 237 struct ino_entry { 238 struct list_head list; /* list head */ 239 nid_t ino; /* inode number */ 240 unsigned int dirty_device; /* dirty device bitmap */ 241 }; 242 243 /* for the list of inodes to be GCed */ 244 struct inode_entry { 245 struct list_head list; /* list head */ 246 struct inode *inode; /* vfs inode pointer */ 247 }; 248 249 struct fsync_node_entry { 250 struct list_head list; /* list head */ 251 struct page *page; /* warm node page pointer */ 252 unsigned int seq_id; /* sequence id */ 253 }; 254 255 /* for the bitmap indicate blocks to be discarded */ 256 struct discard_entry { 257 struct list_head list; /* list head */ 258 block_t start_blkaddr; /* start blockaddr of current segment */ 259 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 260 }; 261 262 /* default discard granularity of inner discard thread, unit: block count */ 263 #define DEFAULT_DISCARD_GRANULARITY 16 264 265 /* max discard pend list number */ 266 #define MAX_PLIST_NUM 512 267 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 268 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 269 270 enum { 271 D_PREP, /* initial */ 272 D_PARTIAL, /* partially submitted */ 273 D_SUBMIT, /* all submitted */ 274 D_DONE, /* finished */ 275 }; 276 277 struct discard_info { 278 block_t lstart; /* logical start address */ 279 block_t len; /* length */ 280 block_t start; /* actual start address in dev */ 281 }; 282 283 struct discard_cmd { 284 struct rb_node rb_node; /* rb node located in rb-tree */ 285 union { 286 struct { 287 block_t lstart; /* logical start address */ 288 block_t len; /* length */ 289 block_t start; /* actual start address in dev */ 290 }; 291 struct discard_info di; /* discard info */ 292 293 }; 294 struct list_head list; /* command list */ 295 struct completion wait; /* compleation */ 296 struct block_device *bdev; /* bdev */ 297 unsigned short ref; /* reference count */ 298 unsigned char state; /* state */ 299 unsigned char queued; /* queued discard */ 300 int error; /* bio error */ 301 spinlock_t lock; /* for state/bio_ref updating */ 302 unsigned short bio_ref; /* bio reference count */ 303 }; 304 305 enum { 306 DPOLICY_BG, 307 DPOLICY_FORCE, 308 DPOLICY_FSTRIM, 309 DPOLICY_UMOUNT, 310 MAX_DPOLICY, 311 }; 312 313 struct discard_policy { 314 int type; /* type of discard */ 315 unsigned int min_interval; /* used for candidates exist */ 316 unsigned int mid_interval; /* used for device busy */ 317 unsigned int max_interval; /* used for candidates not exist */ 318 unsigned int max_requests; /* # of discards issued per round */ 319 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 320 bool io_aware; /* issue discard in idle time */ 321 bool sync; /* submit discard with REQ_SYNC flag */ 322 bool ordered; /* issue discard by lba order */ 323 unsigned int granularity; /* discard granularity */ 324 int timeout; /* discard timeout for put_super */ 325 }; 326 327 struct discard_cmd_control { 328 struct task_struct *f2fs_issue_discard; /* discard thread */ 329 struct list_head entry_list; /* 4KB discard entry list */ 330 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 331 struct list_head wait_list; /* store on-flushing entries */ 332 struct list_head fstrim_list; /* in-flight discard from fstrim */ 333 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 334 unsigned int discard_wake; /* to wake up discard thread */ 335 struct mutex cmd_lock; 336 unsigned int nr_discards; /* # of discards in the list */ 337 unsigned int max_discards; /* max. discards to be issued */ 338 unsigned int discard_granularity; /* discard granularity */ 339 unsigned int undiscard_blks; /* # of undiscard blocks */ 340 unsigned int next_pos; /* next discard position */ 341 atomic_t issued_discard; /* # of issued discard */ 342 atomic_t queued_discard; /* # of queued discard */ 343 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 344 struct rb_root_cached root; /* root of discard rb-tree */ 345 bool rbtree_check; /* config for consistence check */ 346 }; 347 348 /* for the list of fsync inodes, used only during recovery */ 349 struct fsync_inode_entry { 350 struct list_head list; /* list head */ 351 struct inode *inode; /* vfs inode pointer */ 352 block_t blkaddr; /* block address locating the last fsync */ 353 block_t last_dentry; /* block address locating the last dentry */ 354 }; 355 356 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 357 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 358 359 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 360 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 361 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 362 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 363 364 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 365 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 366 367 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 368 { 369 int before = nats_in_cursum(journal); 370 371 journal->n_nats = cpu_to_le16(before + i); 372 return before; 373 } 374 375 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 376 { 377 int before = sits_in_cursum(journal); 378 379 journal->n_sits = cpu_to_le16(before + i); 380 return before; 381 } 382 383 static inline bool __has_cursum_space(struct f2fs_journal *journal, 384 int size, int type) 385 { 386 if (type == NAT_JOURNAL) 387 return size <= MAX_NAT_JENTRIES(journal); 388 return size <= MAX_SIT_JENTRIES(journal); 389 } 390 391 /* 392 * ioctl commands 393 */ 394 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 395 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 396 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 397 398 #define F2FS_IOCTL_MAGIC 0xf5 399 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 400 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 401 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 402 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 403 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 404 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 405 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 406 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 407 struct f2fs_defragment) 408 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 409 struct f2fs_move_range) 410 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 411 struct f2fs_flush_device) 412 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 413 struct f2fs_gc_range) 414 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 415 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 416 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 417 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 418 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 419 420 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 421 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 422 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 423 424 /* 425 * should be same as XFS_IOC_GOINGDOWN. 426 * Flags for going down operation used by FS_IOC_GOINGDOWN 427 */ 428 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 429 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 430 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 431 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 432 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 433 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 434 435 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 436 /* 437 * ioctl commands in 32 bit emulation 438 */ 439 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 440 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 441 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 442 #endif 443 444 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 445 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 446 447 struct f2fs_gc_range { 448 u32 sync; 449 u64 start; 450 u64 len; 451 }; 452 453 struct f2fs_defragment { 454 u64 start; 455 u64 len; 456 }; 457 458 struct f2fs_move_range { 459 u32 dst_fd; /* destination fd */ 460 u64 pos_in; /* start position in src_fd */ 461 u64 pos_out; /* start position in dst_fd */ 462 u64 len; /* size to move */ 463 }; 464 465 struct f2fs_flush_device { 466 u32 dev_num; /* device number to flush */ 467 u32 segments; /* # of segments to flush */ 468 }; 469 470 /* for inline stuff */ 471 #define DEF_INLINE_RESERVED_SIZE 1 472 static inline int get_extra_isize(struct inode *inode); 473 static inline int get_inline_xattr_addrs(struct inode *inode); 474 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 475 (CUR_ADDRS_PER_INODE(inode) - \ 476 get_inline_xattr_addrs(inode) - \ 477 DEF_INLINE_RESERVED_SIZE)) 478 479 /* for inline dir */ 480 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 481 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 482 BITS_PER_BYTE + 1)) 483 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \ 484 BITS_PER_BYTE - 1) / BITS_PER_BYTE) 485 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 486 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 487 NR_INLINE_DENTRY(inode) + \ 488 INLINE_DENTRY_BITMAP_SIZE(inode))) 489 490 /* 491 * For INODE and NODE manager 492 */ 493 /* for directory operations */ 494 struct f2fs_dentry_ptr { 495 struct inode *inode; 496 void *bitmap; 497 struct f2fs_dir_entry *dentry; 498 __u8 (*filename)[F2FS_SLOT_LEN]; 499 int max; 500 int nr_bitmap; 501 }; 502 503 static inline void make_dentry_ptr_block(struct inode *inode, 504 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 505 { 506 d->inode = inode; 507 d->max = NR_DENTRY_IN_BLOCK; 508 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 509 d->bitmap = t->dentry_bitmap; 510 d->dentry = t->dentry; 511 d->filename = t->filename; 512 } 513 514 static inline void make_dentry_ptr_inline(struct inode *inode, 515 struct f2fs_dentry_ptr *d, void *t) 516 { 517 int entry_cnt = NR_INLINE_DENTRY(inode); 518 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 519 int reserved_size = INLINE_RESERVED_SIZE(inode); 520 521 d->inode = inode; 522 d->max = entry_cnt; 523 d->nr_bitmap = bitmap_size; 524 d->bitmap = t; 525 d->dentry = t + bitmap_size + reserved_size; 526 d->filename = t + bitmap_size + reserved_size + 527 SIZE_OF_DIR_ENTRY * entry_cnt; 528 } 529 530 /* 531 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 532 * as its node offset to distinguish from index node blocks. 533 * But some bits are used to mark the node block. 534 */ 535 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 536 >> OFFSET_BIT_SHIFT) 537 enum { 538 ALLOC_NODE, /* allocate a new node page if needed */ 539 LOOKUP_NODE, /* look up a node without readahead */ 540 LOOKUP_NODE_RA, /* 541 * look up a node with readahead called 542 * by get_data_block. 543 */ 544 }; 545 546 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 547 548 /* maximum retry quota flush count */ 549 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 550 551 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 552 553 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 554 555 /* for in-memory extent cache entry */ 556 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 557 558 /* number of extent info in extent cache we try to shrink */ 559 #define EXTENT_CACHE_SHRINK_NUMBER 128 560 561 struct rb_entry { 562 struct rb_node rb_node; /* rb node located in rb-tree */ 563 unsigned int ofs; /* start offset of the entry */ 564 unsigned int len; /* length of the entry */ 565 }; 566 567 struct extent_info { 568 unsigned int fofs; /* start offset in a file */ 569 unsigned int len; /* length of the extent */ 570 u32 blk; /* start block address of the extent */ 571 }; 572 573 struct extent_node { 574 struct rb_node rb_node; /* rb node located in rb-tree */ 575 struct extent_info ei; /* extent info */ 576 struct list_head list; /* node in global extent list of sbi */ 577 struct extent_tree *et; /* extent tree pointer */ 578 }; 579 580 struct extent_tree { 581 nid_t ino; /* inode number */ 582 struct rb_root_cached root; /* root of extent info rb-tree */ 583 struct extent_node *cached_en; /* recently accessed extent node */ 584 struct extent_info largest; /* largested extent info */ 585 struct list_head list; /* to be used by sbi->zombie_list */ 586 rwlock_t lock; /* protect extent info rb-tree */ 587 atomic_t node_cnt; /* # of extent node in rb-tree*/ 588 bool largest_updated; /* largest extent updated */ 589 }; 590 591 /* 592 * This structure is taken from ext4_map_blocks. 593 * 594 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 595 */ 596 #define F2FS_MAP_NEW (1 << BH_New) 597 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 598 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 599 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 600 F2FS_MAP_UNWRITTEN) 601 602 struct f2fs_map_blocks { 603 block_t m_pblk; 604 block_t m_lblk; 605 unsigned int m_len; 606 unsigned int m_flags; 607 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 608 pgoff_t *m_next_extent; /* point to next possible extent */ 609 int m_seg_type; 610 bool m_may_create; /* indicate it is from write path */ 611 }; 612 613 /* for flag in get_data_block */ 614 enum { 615 F2FS_GET_BLOCK_DEFAULT, 616 F2FS_GET_BLOCK_FIEMAP, 617 F2FS_GET_BLOCK_BMAP, 618 F2FS_GET_BLOCK_DIO, 619 F2FS_GET_BLOCK_PRE_DIO, 620 F2FS_GET_BLOCK_PRE_AIO, 621 F2FS_GET_BLOCK_PRECACHE, 622 }; 623 624 /* 625 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 626 */ 627 #define FADVISE_COLD_BIT 0x01 628 #define FADVISE_LOST_PINO_BIT 0x02 629 #define FADVISE_ENCRYPT_BIT 0x04 630 #define FADVISE_ENC_NAME_BIT 0x08 631 #define FADVISE_KEEP_SIZE_BIT 0x10 632 #define FADVISE_HOT_BIT 0x20 633 #define FADVISE_VERITY_BIT 0x40 /* reserved */ 634 635 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 636 637 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 638 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 639 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 640 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 641 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 642 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 643 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 644 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 645 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 646 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 647 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 648 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 649 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 650 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 651 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 652 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 653 654 #define DEF_DIR_LEVEL 0 655 656 enum { 657 GC_FAILURE_PIN, 658 GC_FAILURE_ATOMIC, 659 MAX_GC_FAILURE 660 }; 661 662 struct f2fs_inode_info { 663 struct inode vfs_inode; /* serve a vfs inode */ 664 unsigned long i_flags; /* keep an inode flags for ioctl */ 665 unsigned char i_advise; /* use to give file attribute hints */ 666 unsigned char i_dir_level; /* use for dentry level for large dir */ 667 unsigned int i_current_depth; /* only for directory depth */ 668 /* for gc failure statistic */ 669 unsigned int i_gc_failures[MAX_GC_FAILURE]; 670 unsigned int i_pino; /* parent inode number */ 671 umode_t i_acl_mode; /* keep file acl mode temporarily */ 672 673 /* Use below internally in f2fs*/ 674 unsigned long flags; /* use to pass per-file flags */ 675 struct rw_semaphore i_sem; /* protect fi info */ 676 atomic_t dirty_pages; /* # of dirty pages */ 677 f2fs_hash_t chash; /* hash value of given file name */ 678 unsigned int clevel; /* maximum level of given file name */ 679 struct task_struct *task; /* lookup and create consistency */ 680 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 681 nid_t i_xattr_nid; /* node id that contains xattrs */ 682 loff_t last_disk_size; /* lastly written file size */ 683 684 #ifdef CONFIG_QUOTA 685 struct dquot *i_dquot[MAXQUOTAS]; 686 687 /* quota space reservation, managed internally by quota code */ 688 qsize_t i_reserved_quota; 689 #endif 690 struct list_head dirty_list; /* dirty list for dirs and files */ 691 struct list_head gdirty_list; /* linked in global dirty list */ 692 struct list_head inmem_ilist; /* list for inmem inodes */ 693 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 694 struct task_struct *inmem_task; /* store inmemory task */ 695 struct mutex inmem_lock; /* lock for inmemory pages */ 696 struct extent_tree *extent_tree; /* cached extent_tree entry */ 697 698 /* avoid racing between foreground op and gc */ 699 struct rw_semaphore i_gc_rwsem[2]; 700 struct rw_semaphore i_mmap_sem; 701 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 702 703 int i_extra_isize; /* size of extra space located in i_addr */ 704 kprojid_t i_projid; /* id for project quota */ 705 int i_inline_xattr_size; /* inline xattr size */ 706 struct timespec64 i_crtime; /* inode creation time */ 707 struct timespec64 i_disk_time[4];/* inode disk times */ 708 }; 709 710 static inline void get_extent_info(struct extent_info *ext, 711 struct f2fs_extent *i_ext) 712 { 713 ext->fofs = le32_to_cpu(i_ext->fofs); 714 ext->blk = le32_to_cpu(i_ext->blk); 715 ext->len = le32_to_cpu(i_ext->len); 716 } 717 718 static inline void set_raw_extent(struct extent_info *ext, 719 struct f2fs_extent *i_ext) 720 { 721 i_ext->fofs = cpu_to_le32(ext->fofs); 722 i_ext->blk = cpu_to_le32(ext->blk); 723 i_ext->len = cpu_to_le32(ext->len); 724 } 725 726 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 727 u32 blk, unsigned int len) 728 { 729 ei->fofs = fofs; 730 ei->blk = blk; 731 ei->len = len; 732 } 733 734 static inline bool __is_discard_mergeable(struct discard_info *back, 735 struct discard_info *front, unsigned int max_len) 736 { 737 return (back->lstart + back->len == front->lstart) && 738 (back->len + front->len <= max_len); 739 } 740 741 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 742 struct discard_info *back, unsigned int max_len) 743 { 744 return __is_discard_mergeable(back, cur, max_len); 745 } 746 747 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 748 struct discard_info *front, unsigned int max_len) 749 { 750 return __is_discard_mergeable(cur, front, max_len); 751 } 752 753 static inline bool __is_extent_mergeable(struct extent_info *back, 754 struct extent_info *front) 755 { 756 return (back->fofs + back->len == front->fofs && 757 back->blk + back->len == front->blk); 758 } 759 760 static inline bool __is_back_mergeable(struct extent_info *cur, 761 struct extent_info *back) 762 { 763 return __is_extent_mergeable(back, cur); 764 } 765 766 static inline bool __is_front_mergeable(struct extent_info *cur, 767 struct extent_info *front) 768 { 769 return __is_extent_mergeable(cur, front); 770 } 771 772 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 773 static inline void __try_update_largest_extent(struct extent_tree *et, 774 struct extent_node *en) 775 { 776 if (en->ei.len > et->largest.len) { 777 et->largest = en->ei; 778 et->largest_updated = true; 779 } 780 } 781 782 /* 783 * For free nid management 784 */ 785 enum nid_state { 786 FREE_NID, /* newly added to free nid list */ 787 PREALLOC_NID, /* it is preallocated */ 788 MAX_NID_STATE, 789 }; 790 791 struct f2fs_nm_info { 792 block_t nat_blkaddr; /* base disk address of NAT */ 793 nid_t max_nid; /* maximum possible node ids */ 794 nid_t available_nids; /* # of available node ids */ 795 nid_t next_scan_nid; /* the next nid to be scanned */ 796 unsigned int ram_thresh; /* control the memory footprint */ 797 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 798 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 799 800 /* NAT cache management */ 801 struct radix_tree_root nat_root;/* root of the nat entry cache */ 802 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 803 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 804 struct list_head nat_entries; /* cached nat entry list (clean) */ 805 spinlock_t nat_list_lock; /* protect clean nat entry list */ 806 unsigned int nat_cnt; /* the # of cached nat entries */ 807 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 808 unsigned int nat_blocks; /* # of nat blocks */ 809 810 /* free node ids management */ 811 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 812 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 813 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 814 spinlock_t nid_list_lock; /* protect nid lists ops */ 815 struct mutex build_lock; /* lock for build free nids */ 816 unsigned char **free_nid_bitmap; 817 unsigned char *nat_block_bitmap; 818 unsigned short *free_nid_count; /* free nid count of NAT block */ 819 820 /* for checkpoint */ 821 char *nat_bitmap; /* NAT bitmap pointer */ 822 823 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 824 unsigned char *nat_bits; /* NAT bits blocks */ 825 unsigned char *full_nat_bits; /* full NAT pages */ 826 unsigned char *empty_nat_bits; /* empty NAT pages */ 827 #ifdef CONFIG_F2FS_CHECK_FS 828 char *nat_bitmap_mir; /* NAT bitmap mirror */ 829 #endif 830 int bitmap_size; /* bitmap size */ 831 }; 832 833 /* 834 * this structure is used as one of function parameters. 835 * all the information are dedicated to a given direct node block determined 836 * by the data offset in a file. 837 */ 838 struct dnode_of_data { 839 struct inode *inode; /* vfs inode pointer */ 840 struct page *inode_page; /* its inode page, NULL is possible */ 841 struct page *node_page; /* cached direct node page */ 842 nid_t nid; /* node id of the direct node block */ 843 unsigned int ofs_in_node; /* data offset in the node page */ 844 bool inode_page_locked; /* inode page is locked or not */ 845 bool node_changed; /* is node block changed */ 846 char cur_level; /* level of hole node page */ 847 char max_level; /* level of current page located */ 848 block_t data_blkaddr; /* block address of the node block */ 849 }; 850 851 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 852 struct page *ipage, struct page *npage, nid_t nid) 853 { 854 memset(dn, 0, sizeof(*dn)); 855 dn->inode = inode; 856 dn->inode_page = ipage; 857 dn->node_page = npage; 858 dn->nid = nid; 859 } 860 861 /* 862 * For SIT manager 863 * 864 * By default, there are 6 active log areas across the whole main area. 865 * When considering hot and cold data separation to reduce cleaning overhead, 866 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 867 * respectively. 868 * In the current design, you should not change the numbers intentionally. 869 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 870 * logs individually according to the underlying devices. (default: 6) 871 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 872 * data and 8 for node logs. 873 */ 874 #define NR_CURSEG_DATA_TYPE (3) 875 #define NR_CURSEG_NODE_TYPE (3) 876 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 877 878 enum { 879 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 880 CURSEG_WARM_DATA, /* data blocks */ 881 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 882 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 883 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 884 CURSEG_COLD_NODE, /* indirect node blocks */ 885 NO_CHECK_TYPE, 886 }; 887 888 struct flush_cmd { 889 struct completion wait; 890 struct llist_node llnode; 891 nid_t ino; 892 int ret; 893 }; 894 895 struct flush_cmd_control { 896 struct task_struct *f2fs_issue_flush; /* flush thread */ 897 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 898 atomic_t issued_flush; /* # of issued flushes */ 899 atomic_t queued_flush; /* # of queued flushes */ 900 struct llist_head issue_list; /* list for command issue */ 901 struct llist_node *dispatch_list; /* list for command dispatch */ 902 }; 903 904 struct f2fs_sm_info { 905 struct sit_info *sit_info; /* whole segment information */ 906 struct free_segmap_info *free_info; /* free segment information */ 907 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 908 struct curseg_info *curseg_array; /* active segment information */ 909 910 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 911 912 block_t seg0_blkaddr; /* block address of 0'th segment */ 913 block_t main_blkaddr; /* start block address of main area */ 914 block_t ssa_blkaddr; /* start block address of SSA area */ 915 916 unsigned int segment_count; /* total # of segments */ 917 unsigned int main_segments; /* # of segments in main area */ 918 unsigned int reserved_segments; /* # of reserved segments */ 919 unsigned int ovp_segments; /* # of overprovision segments */ 920 921 /* a threshold to reclaim prefree segments */ 922 unsigned int rec_prefree_segments; 923 924 /* for batched trimming */ 925 unsigned int trim_sections; /* # of sections to trim */ 926 927 struct list_head sit_entry_set; /* sit entry set list */ 928 929 unsigned int ipu_policy; /* in-place-update policy */ 930 unsigned int min_ipu_util; /* in-place-update threshold */ 931 unsigned int min_fsync_blocks; /* threshold for fsync */ 932 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 933 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 934 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 935 936 /* for flush command control */ 937 struct flush_cmd_control *fcc_info; 938 939 /* for discard command control */ 940 struct discard_cmd_control *dcc_info; 941 }; 942 943 /* 944 * For superblock 945 */ 946 /* 947 * COUNT_TYPE for monitoring 948 * 949 * f2fs monitors the number of several block types such as on-writeback, 950 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 951 */ 952 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 953 enum count_type { 954 F2FS_DIRTY_DENTS, 955 F2FS_DIRTY_DATA, 956 F2FS_DIRTY_QDATA, 957 F2FS_DIRTY_NODES, 958 F2FS_DIRTY_META, 959 F2FS_INMEM_PAGES, 960 F2FS_DIRTY_IMETA, 961 F2FS_WB_CP_DATA, 962 F2FS_WB_DATA, 963 F2FS_RD_DATA, 964 F2FS_RD_NODE, 965 F2FS_RD_META, 966 F2FS_DIO_WRITE, 967 F2FS_DIO_READ, 968 NR_COUNT_TYPE, 969 }; 970 971 /* 972 * The below are the page types of bios used in submit_bio(). 973 * The available types are: 974 * DATA User data pages. It operates as async mode. 975 * NODE Node pages. It operates as async mode. 976 * META FS metadata pages such as SIT, NAT, CP. 977 * NR_PAGE_TYPE The number of page types. 978 * META_FLUSH Make sure the previous pages are written 979 * with waiting the bio's completion 980 * ... Only can be used with META. 981 */ 982 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 983 enum page_type { 984 DATA, 985 NODE, 986 META, 987 NR_PAGE_TYPE, 988 META_FLUSH, 989 INMEM, /* the below types are used by tracepoints only. */ 990 INMEM_DROP, 991 INMEM_INVALIDATE, 992 INMEM_REVOKE, 993 IPU, 994 OPU, 995 }; 996 997 enum temp_type { 998 HOT = 0, /* must be zero for meta bio */ 999 WARM, 1000 COLD, 1001 NR_TEMP_TYPE, 1002 }; 1003 1004 enum need_lock_type { 1005 LOCK_REQ = 0, 1006 LOCK_DONE, 1007 LOCK_RETRY, 1008 }; 1009 1010 enum cp_reason_type { 1011 CP_NO_NEEDED, 1012 CP_NON_REGULAR, 1013 CP_HARDLINK, 1014 CP_SB_NEED_CP, 1015 CP_WRONG_PINO, 1016 CP_NO_SPC_ROLL, 1017 CP_NODE_NEED_CP, 1018 CP_FASTBOOT_MODE, 1019 CP_SPEC_LOG_NUM, 1020 CP_RECOVER_DIR, 1021 }; 1022 1023 enum iostat_type { 1024 APP_DIRECT_IO, /* app direct IOs */ 1025 APP_BUFFERED_IO, /* app buffered IOs */ 1026 APP_WRITE_IO, /* app write IOs */ 1027 APP_MAPPED_IO, /* app mapped IOs */ 1028 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1029 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1030 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1031 FS_GC_DATA_IO, /* data IOs from forground gc */ 1032 FS_GC_NODE_IO, /* node IOs from forground gc */ 1033 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1034 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1035 FS_CP_META_IO, /* meta IOs from checkpoint */ 1036 FS_DISCARD, /* discard */ 1037 NR_IO_TYPE, 1038 }; 1039 1040 struct f2fs_io_info { 1041 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1042 nid_t ino; /* inode number */ 1043 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1044 enum temp_type temp; /* contains HOT/WARM/COLD */ 1045 int op; /* contains REQ_OP_ */ 1046 int op_flags; /* req_flag_bits */ 1047 block_t new_blkaddr; /* new block address to be written */ 1048 block_t old_blkaddr; /* old block address before Cow */ 1049 struct page *page; /* page to be written */ 1050 struct page *encrypted_page; /* encrypted page */ 1051 struct list_head list; /* serialize IOs */ 1052 bool submitted; /* indicate IO submission */ 1053 int need_lock; /* indicate we need to lock cp_rwsem */ 1054 bool in_list; /* indicate fio is in io_list */ 1055 bool is_por; /* indicate IO is from recovery or not */ 1056 bool retry; /* need to reallocate block address */ 1057 enum iostat_type io_type; /* io type */ 1058 struct writeback_control *io_wbc; /* writeback control */ 1059 struct bio **bio; /* bio for ipu */ 1060 sector_t *last_block; /* last block number in bio */ 1061 unsigned char version; /* version of the node */ 1062 }; 1063 1064 #define is_read_io(rw) ((rw) == READ) 1065 struct f2fs_bio_info { 1066 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1067 struct bio *bio; /* bios to merge */ 1068 sector_t last_block_in_bio; /* last block number */ 1069 struct f2fs_io_info fio; /* store buffered io info. */ 1070 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1071 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1072 struct list_head io_list; /* track fios */ 1073 }; 1074 1075 #define FDEV(i) (sbi->devs[i]) 1076 #define RDEV(i) (raw_super->devs[i]) 1077 struct f2fs_dev_info { 1078 struct block_device *bdev; 1079 char path[MAX_PATH_LEN]; 1080 unsigned int total_segments; 1081 block_t start_blk; 1082 block_t end_blk; 1083 #ifdef CONFIG_BLK_DEV_ZONED 1084 unsigned int nr_blkz; /* Total number of zones */ 1085 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1086 #endif 1087 }; 1088 1089 enum inode_type { 1090 DIR_INODE, /* for dirty dir inode */ 1091 FILE_INODE, /* for dirty regular/symlink inode */ 1092 DIRTY_META, /* for all dirtied inode metadata */ 1093 ATOMIC_FILE, /* for all atomic files */ 1094 NR_INODE_TYPE, 1095 }; 1096 1097 /* for inner inode cache management */ 1098 struct inode_management { 1099 struct radix_tree_root ino_root; /* ino entry array */ 1100 spinlock_t ino_lock; /* for ino entry lock */ 1101 struct list_head ino_list; /* inode list head */ 1102 unsigned long ino_num; /* number of entries */ 1103 }; 1104 1105 /* For s_flag in struct f2fs_sb_info */ 1106 enum { 1107 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1108 SBI_IS_CLOSE, /* specify unmounting */ 1109 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1110 SBI_POR_DOING, /* recovery is doing or not */ 1111 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1112 SBI_NEED_CP, /* need to checkpoint */ 1113 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1114 SBI_IS_RECOVERED, /* recovered orphan/data */ 1115 SBI_CP_DISABLED, /* CP was disabled last mount */ 1116 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1117 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1118 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1119 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1120 SBI_IS_RESIZEFS, /* resizefs is in process */ 1121 }; 1122 1123 enum { 1124 CP_TIME, 1125 REQ_TIME, 1126 DISCARD_TIME, 1127 GC_TIME, 1128 DISABLE_TIME, 1129 UMOUNT_DISCARD_TIMEOUT, 1130 MAX_TIME, 1131 }; 1132 1133 enum { 1134 GC_NORMAL, 1135 GC_IDLE_CB, 1136 GC_IDLE_GREEDY, 1137 GC_URGENT, 1138 }; 1139 1140 enum { 1141 WHINT_MODE_OFF, /* not pass down write hints */ 1142 WHINT_MODE_USER, /* try to pass down hints given by users */ 1143 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1144 }; 1145 1146 enum { 1147 ALLOC_MODE_DEFAULT, /* stay default */ 1148 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1149 }; 1150 1151 enum fsync_mode { 1152 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1153 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1154 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1155 }; 1156 1157 #ifdef CONFIG_FS_ENCRYPTION 1158 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1159 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1160 #else 1161 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1162 #endif 1163 1164 struct f2fs_sb_info { 1165 struct super_block *sb; /* pointer to VFS super block */ 1166 struct proc_dir_entry *s_proc; /* proc entry */ 1167 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1168 struct rw_semaphore sb_lock; /* lock for raw super block */ 1169 int valid_super_block; /* valid super block no */ 1170 unsigned long s_flag; /* flags for sbi */ 1171 struct mutex writepages; /* mutex for writepages() */ 1172 1173 #ifdef CONFIG_BLK_DEV_ZONED 1174 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1175 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1176 #endif 1177 1178 /* for node-related operations */ 1179 struct f2fs_nm_info *nm_info; /* node manager */ 1180 struct inode *node_inode; /* cache node blocks */ 1181 1182 /* for segment-related operations */ 1183 struct f2fs_sm_info *sm_info; /* segment manager */ 1184 1185 /* for bio operations */ 1186 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1187 /* keep migration IO order for LFS mode */ 1188 struct rw_semaphore io_order_lock; 1189 mempool_t *write_io_dummy; /* Dummy pages */ 1190 1191 /* for checkpoint */ 1192 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1193 int cur_cp_pack; /* remain current cp pack */ 1194 spinlock_t cp_lock; /* for flag in ckpt */ 1195 struct inode *meta_inode; /* cache meta blocks */ 1196 struct mutex cp_mutex; /* checkpoint procedure lock */ 1197 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1198 struct rw_semaphore node_write; /* locking node writes */ 1199 struct rw_semaphore node_change; /* locking node change */ 1200 wait_queue_head_t cp_wait; 1201 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1202 long interval_time[MAX_TIME]; /* to store thresholds */ 1203 1204 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1205 1206 spinlock_t fsync_node_lock; /* for node entry lock */ 1207 struct list_head fsync_node_list; /* node list head */ 1208 unsigned int fsync_seg_id; /* sequence id */ 1209 unsigned int fsync_node_num; /* number of node entries */ 1210 1211 /* for orphan inode, use 0'th array */ 1212 unsigned int max_orphans; /* max orphan inodes */ 1213 1214 /* for inode management */ 1215 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1216 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1217 struct mutex flush_lock; /* for flush exclusion */ 1218 1219 /* for extent tree cache */ 1220 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1221 struct mutex extent_tree_lock; /* locking extent radix tree */ 1222 struct list_head extent_list; /* lru list for shrinker */ 1223 spinlock_t extent_lock; /* locking extent lru list */ 1224 atomic_t total_ext_tree; /* extent tree count */ 1225 struct list_head zombie_list; /* extent zombie tree list */ 1226 atomic_t total_zombie_tree; /* extent zombie tree count */ 1227 atomic_t total_ext_node; /* extent info count */ 1228 1229 /* basic filesystem units */ 1230 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1231 unsigned int log_blocksize; /* log2 block size */ 1232 unsigned int blocksize; /* block size */ 1233 unsigned int root_ino_num; /* root inode number*/ 1234 unsigned int node_ino_num; /* node inode number*/ 1235 unsigned int meta_ino_num; /* meta inode number*/ 1236 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1237 unsigned int blocks_per_seg; /* blocks per segment */ 1238 unsigned int segs_per_sec; /* segments per section */ 1239 unsigned int secs_per_zone; /* sections per zone */ 1240 unsigned int total_sections; /* total section count */ 1241 struct mutex resize_mutex; /* for resize exclusion */ 1242 unsigned int total_node_count; /* total node block count */ 1243 unsigned int total_valid_node_count; /* valid node block count */ 1244 loff_t max_file_blocks; /* max block index of file */ 1245 int dir_level; /* directory level */ 1246 int readdir_ra; /* readahead inode in readdir */ 1247 1248 block_t user_block_count; /* # of user blocks */ 1249 block_t total_valid_block_count; /* # of valid blocks */ 1250 block_t discard_blks; /* discard command candidats */ 1251 block_t last_valid_block_count; /* for recovery */ 1252 block_t reserved_blocks; /* configurable reserved blocks */ 1253 block_t current_reserved_blocks; /* current reserved blocks */ 1254 1255 /* Additional tracking for no checkpoint mode */ 1256 block_t unusable_block_count; /* # of blocks saved by last cp */ 1257 1258 unsigned int nquota_files; /* # of quota sysfile */ 1259 1260 /* # of pages, see count_type */ 1261 atomic_t nr_pages[NR_COUNT_TYPE]; 1262 /* # of allocated blocks */ 1263 struct percpu_counter alloc_valid_block_count; 1264 1265 /* writeback control */ 1266 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1267 1268 /* valid inode count */ 1269 struct percpu_counter total_valid_inode_count; 1270 1271 struct f2fs_mount_info mount_opt; /* mount options */ 1272 1273 /* for cleaning operations */ 1274 struct mutex gc_mutex; /* mutex for GC */ 1275 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1276 unsigned int cur_victim_sec; /* current victim section num */ 1277 unsigned int gc_mode; /* current GC state */ 1278 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1279 /* for skip statistic */ 1280 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1281 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1282 1283 /* threshold for gc trials on pinned files */ 1284 u64 gc_pin_file_threshold; 1285 1286 /* maximum # of trials to find a victim segment for SSR and GC */ 1287 unsigned int max_victim_search; 1288 /* migration granularity of garbage collection, unit: segment */ 1289 unsigned int migration_granularity; 1290 1291 /* 1292 * for stat information. 1293 * one is for the LFS mode, and the other is for the SSR mode. 1294 */ 1295 #ifdef CONFIG_F2FS_STAT_FS 1296 struct f2fs_stat_info *stat_info; /* FS status information */ 1297 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1298 unsigned int segment_count[2]; /* # of allocated segments */ 1299 unsigned int block_count[2]; /* # of allocated blocks */ 1300 atomic_t inplace_count; /* # of inplace update */ 1301 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1302 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1303 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1304 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1305 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1306 atomic_t inline_inode; /* # of inline_data inodes */ 1307 atomic_t inline_dir; /* # of inline_dentry inodes */ 1308 atomic_t aw_cnt; /* # of atomic writes */ 1309 atomic_t vw_cnt; /* # of volatile writes */ 1310 atomic_t max_aw_cnt; /* max # of atomic writes */ 1311 atomic_t max_vw_cnt; /* max # of volatile writes */ 1312 int bg_gc; /* background gc calls */ 1313 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1314 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1315 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1316 #endif 1317 spinlock_t stat_lock; /* lock for stat operations */ 1318 1319 /* For app/fs IO statistics */ 1320 spinlock_t iostat_lock; 1321 unsigned long long write_iostat[NR_IO_TYPE]; 1322 bool iostat_enable; 1323 1324 /* For sysfs suppport */ 1325 struct kobject s_kobj; 1326 struct completion s_kobj_unregister; 1327 1328 /* For shrinker support */ 1329 struct list_head s_list; 1330 int s_ndevs; /* number of devices */ 1331 struct f2fs_dev_info *devs; /* for device list */ 1332 unsigned int dirty_device; /* for checkpoint data flush */ 1333 spinlock_t dev_lock; /* protect dirty_device */ 1334 struct mutex umount_mutex; 1335 unsigned int shrinker_run_no; 1336 1337 /* For write statistics */ 1338 u64 sectors_written_start; 1339 u64 kbytes_written; 1340 1341 /* Reference to checksum algorithm driver via cryptoapi */ 1342 struct crypto_shash *s_chksum_driver; 1343 1344 /* Precomputed FS UUID checksum for seeding other checksums */ 1345 __u32 s_chksum_seed; 1346 }; 1347 1348 struct f2fs_private_dio { 1349 struct inode *inode; 1350 void *orig_private; 1351 bio_end_io_t *orig_end_io; 1352 bool write; 1353 }; 1354 1355 #ifdef CONFIG_F2FS_FAULT_INJECTION 1356 #define f2fs_show_injection_info(type) \ 1357 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pS\n", \ 1358 KERN_INFO, f2fs_fault_name[type], \ 1359 __func__, __builtin_return_address(0)) 1360 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1361 { 1362 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1363 1364 if (!ffi->inject_rate) 1365 return false; 1366 1367 if (!IS_FAULT_SET(ffi, type)) 1368 return false; 1369 1370 atomic_inc(&ffi->inject_ops); 1371 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1372 atomic_set(&ffi->inject_ops, 0); 1373 return true; 1374 } 1375 return false; 1376 } 1377 #else 1378 #define f2fs_show_injection_info(type) do { } while (0) 1379 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1380 { 1381 return false; 1382 } 1383 #endif 1384 1385 /* 1386 * Test if the mounted volume is a multi-device volume. 1387 * - For a single regular disk volume, sbi->s_ndevs is 0. 1388 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1389 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1390 */ 1391 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1392 { 1393 return sbi->s_ndevs > 1; 1394 } 1395 1396 /* For write statistics. Suppose sector size is 512 bytes, 1397 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1398 */ 1399 #define BD_PART_WRITTEN(s) \ 1400 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1401 (s)->sectors_written_start) >> 1) 1402 1403 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1404 { 1405 unsigned long now = jiffies; 1406 1407 sbi->last_time[type] = now; 1408 1409 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1410 if (type == REQ_TIME) { 1411 sbi->last_time[DISCARD_TIME] = now; 1412 sbi->last_time[GC_TIME] = now; 1413 } 1414 } 1415 1416 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1417 { 1418 unsigned long interval = sbi->interval_time[type] * HZ; 1419 1420 return time_after(jiffies, sbi->last_time[type] + interval); 1421 } 1422 1423 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1424 int type) 1425 { 1426 unsigned long interval = sbi->interval_time[type] * HZ; 1427 unsigned int wait_ms = 0; 1428 long delta; 1429 1430 delta = (sbi->last_time[type] + interval) - jiffies; 1431 if (delta > 0) 1432 wait_ms = jiffies_to_msecs(delta); 1433 1434 return wait_ms; 1435 } 1436 1437 /* 1438 * Inline functions 1439 */ 1440 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1441 const void *address, unsigned int length) 1442 { 1443 struct { 1444 struct shash_desc shash; 1445 char ctx[4]; 1446 } desc; 1447 int err; 1448 1449 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1450 1451 desc.shash.tfm = sbi->s_chksum_driver; 1452 *(u32 *)desc.ctx = crc; 1453 1454 err = crypto_shash_update(&desc.shash, address, length); 1455 BUG_ON(err); 1456 1457 return *(u32 *)desc.ctx; 1458 } 1459 1460 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1461 unsigned int length) 1462 { 1463 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1464 } 1465 1466 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1467 void *buf, size_t buf_size) 1468 { 1469 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1470 } 1471 1472 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1473 const void *address, unsigned int length) 1474 { 1475 return __f2fs_crc32(sbi, crc, address, length); 1476 } 1477 1478 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1479 { 1480 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1481 } 1482 1483 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1484 { 1485 return sb->s_fs_info; 1486 } 1487 1488 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1489 { 1490 return F2FS_SB(inode->i_sb); 1491 } 1492 1493 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1494 { 1495 return F2FS_I_SB(mapping->host); 1496 } 1497 1498 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1499 { 1500 return F2FS_M_SB(page->mapping); 1501 } 1502 1503 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1504 { 1505 return (struct f2fs_super_block *)(sbi->raw_super); 1506 } 1507 1508 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1509 { 1510 return (struct f2fs_checkpoint *)(sbi->ckpt); 1511 } 1512 1513 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1514 { 1515 return (struct f2fs_node *)page_address(page); 1516 } 1517 1518 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1519 { 1520 return &((struct f2fs_node *)page_address(page))->i; 1521 } 1522 1523 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1524 { 1525 return (struct f2fs_nm_info *)(sbi->nm_info); 1526 } 1527 1528 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1529 { 1530 return (struct f2fs_sm_info *)(sbi->sm_info); 1531 } 1532 1533 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1534 { 1535 return (struct sit_info *)(SM_I(sbi)->sit_info); 1536 } 1537 1538 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1539 { 1540 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1541 } 1542 1543 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1544 { 1545 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1546 } 1547 1548 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1549 { 1550 return sbi->meta_inode->i_mapping; 1551 } 1552 1553 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1554 { 1555 return sbi->node_inode->i_mapping; 1556 } 1557 1558 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1559 { 1560 return test_bit(type, &sbi->s_flag); 1561 } 1562 1563 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1564 { 1565 set_bit(type, &sbi->s_flag); 1566 } 1567 1568 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1569 { 1570 clear_bit(type, &sbi->s_flag); 1571 } 1572 1573 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1574 { 1575 return le64_to_cpu(cp->checkpoint_ver); 1576 } 1577 1578 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1579 { 1580 if (type < F2FS_MAX_QUOTAS) 1581 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1582 return 0; 1583 } 1584 1585 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1586 { 1587 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1588 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1589 } 1590 1591 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1592 { 1593 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1594 1595 return ckpt_flags & f; 1596 } 1597 1598 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1599 { 1600 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1601 } 1602 1603 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1604 { 1605 unsigned int ckpt_flags; 1606 1607 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1608 ckpt_flags |= f; 1609 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1610 } 1611 1612 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1613 { 1614 unsigned long flags; 1615 1616 spin_lock_irqsave(&sbi->cp_lock, flags); 1617 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1618 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1619 } 1620 1621 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1622 { 1623 unsigned int ckpt_flags; 1624 1625 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1626 ckpt_flags &= (~f); 1627 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1628 } 1629 1630 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1631 { 1632 unsigned long flags; 1633 1634 spin_lock_irqsave(&sbi->cp_lock, flags); 1635 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1636 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1637 } 1638 1639 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1640 { 1641 unsigned long flags; 1642 1643 /* 1644 * In order to re-enable nat_bits we need to call fsck.f2fs by 1645 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1646 * so let's rely on regular fsck or unclean shutdown. 1647 */ 1648 1649 if (lock) 1650 spin_lock_irqsave(&sbi->cp_lock, flags); 1651 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1652 kvfree(NM_I(sbi)->nat_bits); 1653 NM_I(sbi)->nat_bits = NULL; 1654 if (lock) 1655 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1656 } 1657 1658 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1659 struct cp_control *cpc) 1660 { 1661 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1662 1663 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1664 } 1665 1666 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1667 { 1668 down_read(&sbi->cp_rwsem); 1669 } 1670 1671 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1672 { 1673 return down_read_trylock(&sbi->cp_rwsem); 1674 } 1675 1676 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1677 { 1678 up_read(&sbi->cp_rwsem); 1679 } 1680 1681 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1682 { 1683 down_write(&sbi->cp_rwsem); 1684 } 1685 1686 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1687 { 1688 up_write(&sbi->cp_rwsem); 1689 } 1690 1691 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1692 { 1693 int reason = CP_SYNC; 1694 1695 if (test_opt(sbi, FASTBOOT)) 1696 reason = CP_FASTBOOT; 1697 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1698 reason = CP_UMOUNT; 1699 return reason; 1700 } 1701 1702 static inline bool __remain_node_summaries(int reason) 1703 { 1704 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1705 } 1706 1707 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1708 { 1709 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1710 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1711 } 1712 1713 /* 1714 * Check whether the inode has blocks or not 1715 */ 1716 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1717 { 1718 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1719 1720 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1721 } 1722 1723 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1724 { 1725 return ofs == XATTR_NODE_OFFSET; 1726 } 1727 1728 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1729 struct inode *inode, bool cap) 1730 { 1731 if (!inode) 1732 return true; 1733 if (!test_opt(sbi, RESERVE_ROOT)) 1734 return false; 1735 if (IS_NOQUOTA(inode)) 1736 return true; 1737 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1738 return true; 1739 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1740 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1741 return true; 1742 if (cap && capable(CAP_SYS_RESOURCE)) 1743 return true; 1744 return false; 1745 } 1746 1747 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1748 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1749 struct inode *inode, blkcnt_t *count) 1750 { 1751 blkcnt_t diff = 0, release = 0; 1752 block_t avail_user_block_count; 1753 int ret; 1754 1755 ret = dquot_reserve_block(inode, *count); 1756 if (ret) 1757 return ret; 1758 1759 if (time_to_inject(sbi, FAULT_BLOCK)) { 1760 f2fs_show_injection_info(FAULT_BLOCK); 1761 release = *count; 1762 goto enospc; 1763 } 1764 1765 /* 1766 * let's increase this in prior to actual block count change in order 1767 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1768 */ 1769 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1770 1771 spin_lock(&sbi->stat_lock); 1772 sbi->total_valid_block_count += (block_t)(*count); 1773 avail_user_block_count = sbi->user_block_count - 1774 sbi->current_reserved_blocks; 1775 1776 if (!__allow_reserved_blocks(sbi, inode, true)) 1777 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1778 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 1779 if (avail_user_block_count > sbi->unusable_block_count) 1780 avail_user_block_count -= sbi->unusable_block_count; 1781 else 1782 avail_user_block_count = 0; 1783 } 1784 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1785 diff = sbi->total_valid_block_count - avail_user_block_count; 1786 if (diff > *count) 1787 diff = *count; 1788 *count -= diff; 1789 release = diff; 1790 sbi->total_valid_block_count -= diff; 1791 if (!*count) { 1792 spin_unlock(&sbi->stat_lock); 1793 goto enospc; 1794 } 1795 } 1796 spin_unlock(&sbi->stat_lock); 1797 1798 if (unlikely(release)) { 1799 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1800 dquot_release_reservation_block(inode, release); 1801 } 1802 f2fs_i_blocks_write(inode, *count, true, true); 1803 return 0; 1804 1805 enospc: 1806 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1807 dquot_release_reservation_block(inode, release); 1808 return -ENOSPC; 1809 } 1810 1811 __printf(2, 3) 1812 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 1813 1814 #define f2fs_err(sbi, fmt, ...) \ 1815 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 1816 #define f2fs_warn(sbi, fmt, ...) \ 1817 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 1818 #define f2fs_notice(sbi, fmt, ...) \ 1819 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 1820 #define f2fs_info(sbi, fmt, ...) \ 1821 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 1822 #define f2fs_debug(sbi, fmt, ...) \ 1823 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 1824 1825 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 1826 struct inode *inode, 1827 block_t count) 1828 { 1829 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 1830 1831 spin_lock(&sbi->stat_lock); 1832 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 1833 sbi->total_valid_block_count -= (block_t)count; 1834 if (sbi->reserved_blocks && 1835 sbi->current_reserved_blocks < sbi->reserved_blocks) 1836 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 1837 sbi->current_reserved_blocks + count); 1838 spin_unlock(&sbi->stat_lock); 1839 if (unlikely(inode->i_blocks < sectors)) { 1840 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 1841 inode->i_ino, 1842 (unsigned long long)inode->i_blocks, 1843 (unsigned long long)sectors); 1844 set_sbi_flag(sbi, SBI_NEED_FSCK); 1845 return; 1846 } 1847 f2fs_i_blocks_write(inode, count, false, true); 1848 } 1849 1850 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 1851 { 1852 atomic_inc(&sbi->nr_pages[count_type]); 1853 1854 if (count_type == F2FS_DIRTY_DENTS || 1855 count_type == F2FS_DIRTY_NODES || 1856 count_type == F2FS_DIRTY_META || 1857 count_type == F2FS_DIRTY_QDATA || 1858 count_type == F2FS_DIRTY_IMETA) 1859 set_sbi_flag(sbi, SBI_IS_DIRTY); 1860 } 1861 1862 static inline void inode_inc_dirty_pages(struct inode *inode) 1863 { 1864 atomic_inc(&F2FS_I(inode)->dirty_pages); 1865 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1866 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1867 if (IS_NOQUOTA(inode)) 1868 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1869 } 1870 1871 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 1872 { 1873 atomic_dec(&sbi->nr_pages[count_type]); 1874 } 1875 1876 static inline void inode_dec_dirty_pages(struct inode *inode) 1877 { 1878 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 1879 !S_ISLNK(inode->i_mode)) 1880 return; 1881 1882 atomic_dec(&F2FS_I(inode)->dirty_pages); 1883 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1884 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1885 if (IS_NOQUOTA(inode)) 1886 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1887 } 1888 1889 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 1890 { 1891 return atomic_read(&sbi->nr_pages[count_type]); 1892 } 1893 1894 static inline int get_dirty_pages(struct inode *inode) 1895 { 1896 return atomic_read(&F2FS_I(inode)->dirty_pages); 1897 } 1898 1899 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 1900 { 1901 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 1902 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 1903 sbi->log_blocks_per_seg; 1904 1905 return segs / sbi->segs_per_sec; 1906 } 1907 1908 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 1909 { 1910 return sbi->total_valid_block_count; 1911 } 1912 1913 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 1914 { 1915 return sbi->discard_blks; 1916 } 1917 1918 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 1919 { 1920 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1921 1922 /* return NAT or SIT bitmap */ 1923 if (flag == NAT_BITMAP) 1924 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 1925 else if (flag == SIT_BITMAP) 1926 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 1927 1928 return 0; 1929 } 1930 1931 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 1932 { 1933 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 1934 } 1935 1936 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 1937 { 1938 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1939 int offset; 1940 1941 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 1942 offset = (flag == SIT_BITMAP) ? 1943 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 1944 /* 1945 * if large_nat_bitmap feature is enabled, leave checksum 1946 * protection for all nat/sit bitmaps. 1947 */ 1948 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 1949 } 1950 1951 if (__cp_payload(sbi) > 0) { 1952 if (flag == NAT_BITMAP) 1953 return &ckpt->sit_nat_version_bitmap; 1954 else 1955 return (unsigned char *)ckpt + F2FS_BLKSIZE; 1956 } else { 1957 offset = (flag == NAT_BITMAP) ? 1958 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 1959 return &ckpt->sit_nat_version_bitmap + offset; 1960 } 1961 } 1962 1963 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 1964 { 1965 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1966 1967 if (sbi->cur_cp_pack == 2) 1968 start_addr += sbi->blocks_per_seg; 1969 return start_addr; 1970 } 1971 1972 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 1973 { 1974 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1975 1976 if (sbi->cur_cp_pack == 1) 1977 start_addr += sbi->blocks_per_seg; 1978 return start_addr; 1979 } 1980 1981 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 1982 { 1983 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 1984 } 1985 1986 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 1987 { 1988 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 1989 } 1990 1991 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 1992 struct inode *inode, bool is_inode) 1993 { 1994 block_t valid_block_count; 1995 unsigned int valid_node_count, user_block_count; 1996 int err; 1997 1998 if (is_inode) { 1999 if (inode) { 2000 err = dquot_alloc_inode(inode); 2001 if (err) 2002 return err; 2003 } 2004 } else { 2005 err = dquot_reserve_block(inode, 1); 2006 if (err) 2007 return err; 2008 } 2009 2010 if (time_to_inject(sbi, FAULT_BLOCK)) { 2011 f2fs_show_injection_info(FAULT_BLOCK); 2012 goto enospc; 2013 } 2014 2015 spin_lock(&sbi->stat_lock); 2016 2017 valid_block_count = sbi->total_valid_block_count + 2018 sbi->current_reserved_blocks + 1; 2019 2020 if (!__allow_reserved_blocks(sbi, inode, false)) 2021 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2022 user_block_count = sbi->user_block_count; 2023 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2024 user_block_count -= sbi->unusable_block_count; 2025 2026 if (unlikely(valid_block_count > user_block_count)) { 2027 spin_unlock(&sbi->stat_lock); 2028 goto enospc; 2029 } 2030 2031 valid_node_count = sbi->total_valid_node_count + 1; 2032 if (unlikely(valid_node_count > sbi->total_node_count)) { 2033 spin_unlock(&sbi->stat_lock); 2034 goto enospc; 2035 } 2036 2037 sbi->total_valid_node_count++; 2038 sbi->total_valid_block_count++; 2039 spin_unlock(&sbi->stat_lock); 2040 2041 if (inode) { 2042 if (is_inode) 2043 f2fs_mark_inode_dirty_sync(inode, true); 2044 else 2045 f2fs_i_blocks_write(inode, 1, true, true); 2046 } 2047 2048 percpu_counter_inc(&sbi->alloc_valid_block_count); 2049 return 0; 2050 2051 enospc: 2052 if (is_inode) { 2053 if (inode) 2054 dquot_free_inode(inode); 2055 } else { 2056 dquot_release_reservation_block(inode, 1); 2057 } 2058 return -ENOSPC; 2059 } 2060 2061 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2062 struct inode *inode, bool is_inode) 2063 { 2064 spin_lock(&sbi->stat_lock); 2065 2066 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2067 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2068 2069 sbi->total_valid_node_count--; 2070 sbi->total_valid_block_count--; 2071 if (sbi->reserved_blocks && 2072 sbi->current_reserved_blocks < sbi->reserved_blocks) 2073 sbi->current_reserved_blocks++; 2074 2075 spin_unlock(&sbi->stat_lock); 2076 2077 if (is_inode) { 2078 dquot_free_inode(inode); 2079 } else { 2080 if (unlikely(inode->i_blocks == 0)) { 2081 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu", 2082 inode->i_ino, 2083 (unsigned long long)inode->i_blocks); 2084 set_sbi_flag(sbi, SBI_NEED_FSCK); 2085 return; 2086 } 2087 f2fs_i_blocks_write(inode, 1, false, true); 2088 } 2089 } 2090 2091 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2092 { 2093 return sbi->total_valid_node_count; 2094 } 2095 2096 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2097 { 2098 percpu_counter_inc(&sbi->total_valid_inode_count); 2099 } 2100 2101 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2102 { 2103 percpu_counter_dec(&sbi->total_valid_inode_count); 2104 } 2105 2106 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2107 { 2108 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2109 } 2110 2111 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2112 pgoff_t index, bool for_write) 2113 { 2114 struct page *page; 2115 2116 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2117 if (!for_write) 2118 page = find_get_page_flags(mapping, index, 2119 FGP_LOCK | FGP_ACCESSED); 2120 else 2121 page = find_lock_page(mapping, index); 2122 if (page) 2123 return page; 2124 2125 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2126 f2fs_show_injection_info(FAULT_PAGE_ALLOC); 2127 return NULL; 2128 } 2129 } 2130 2131 if (!for_write) 2132 return grab_cache_page(mapping, index); 2133 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2134 } 2135 2136 static inline struct page *f2fs_pagecache_get_page( 2137 struct address_space *mapping, pgoff_t index, 2138 int fgp_flags, gfp_t gfp_mask) 2139 { 2140 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2141 f2fs_show_injection_info(FAULT_PAGE_GET); 2142 return NULL; 2143 } 2144 2145 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2146 } 2147 2148 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2149 { 2150 char *src_kaddr = kmap(src); 2151 char *dst_kaddr = kmap(dst); 2152 2153 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2154 kunmap(dst); 2155 kunmap(src); 2156 } 2157 2158 static inline void f2fs_put_page(struct page *page, int unlock) 2159 { 2160 if (!page) 2161 return; 2162 2163 if (unlock) { 2164 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2165 unlock_page(page); 2166 } 2167 put_page(page); 2168 } 2169 2170 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2171 { 2172 if (dn->node_page) 2173 f2fs_put_page(dn->node_page, 1); 2174 if (dn->inode_page && dn->node_page != dn->inode_page) 2175 f2fs_put_page(dn->inode_page, 0); 2176 dn->node_page = NULL; 2177 dn->inode_page = NULL; 2178 } 2179 2180 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2181 size_t size) 2182 { 2183 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2184 } 2185 2186 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2187 gfp_t flags) 2188 { 2189 void *entry; 2190 2191 entry = kmem_cache_alloc(cachep, flags); 2192 if (!entry) 2193 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2194 return entry; 2195 } 2196 2197 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, 2198 int npages, bool no_fail) 2199 { 2200 struct bio *bio; 2201 2202 if (no_fail) { 2203 /* No failure on bio allocation */ 2204 bio = bio_alloc(GFP_NOIO, npages); 2205 if (!bio) 2206 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages); 2207 return bio; 2208 } 2209 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) { 2210 f2fs_show_injection_info(FAULT_ALLOC_BIO); 2211 return NULL; 2212 } 2213 2214 return bio_alloc(GFP_KERNEL, npages); 2215 } 2216 2217 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2218 { 2219 if (sbi->gc_mode == GC_URGENT) 2220 return true; 2221 2222 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2223 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2224 get_pages(sbi, F2FS_WB_CP_DATA) || 2225 get_pages(sbi, F2FS_DIO_READ) || 2226 get_pages(sbi, F2FS_DIO_WRITE)) 2227 return false; 2228 2229 if (SM_I(sbi) && SM_I(sbi)->dcc_info && 2230 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2231 return false; 2232 2233 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2234 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2235 return false; 2236 2237 return f2fs_time_over(sbi, type); 2238 } 2239 2240 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2241 unsigned long index, void *item) 2242 { 2243 while (radix_tree_insert(root, index, item)) 2244 cond_resched(); 2245 } 2246 2247 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2248 2249 static inline bool IS_INODE(struct page *page) 2250 { 2251 struct f2fs_node *p = F2FS_NODE(page); 2252 2253 return RAW_IS_INODE(p); 2254 } 2255 2256 static inline int offset_in_addr(struct f2fs_inode *i) 2257 { 2258 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2259 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2260 } 2261 2262 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2263 { 2264 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2265 } 2266 2267 static inline int f2fs_has_extra_attr(struct inode *inode); 2268 static inline block_t datablock_addr(struct inode *inode, 2269 struct page *node_page, unsigned int offset) 2270 { 2271 struct f2fs_node *raw_node; 2272 __le32 *addr_array; 2273 int base = 0; 2274 bool is_inode = IS_INODE(node_page); 2275 2276 raw_node = F2FS_NODE(node_page); 2277 2278 /* from GC path only */ 2279 if (is_inode) { 2280 if (!inode) 2281 base = offset_in_addr(&raw_node->i); 2282 else if (f2fs_has_extra_attr(inode)) 2283 base = get_extra_isize(inode); 2284 } 2285 2286 addr_array = blkaddr_in_node(raw_node); 2287 return le32_to_cpu(addr_array[base + offset]); 2288 } 2289 2290 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2291 { 2292 int mask; 2293 2294 addr += (nr >> 3); 2295 mask = 1 << (7 - (nr & 0x07)); 2296 return mask & *addr; 2297 } 2298 2299 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2300 { 2301 int mask; 2302 2303 addr += (nr >> 3); 2304 mask = 1 << (7 - (nr & 0x07)); 2305 *addr |= mask; 2306 } 2307 2308 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2309 { 2310 int mask; 2311 2312 addr += (nr >> 3); 2313 mask = 1 << (7 - (nr & 0x07)); 2314 *addr &= ~mask; 2315 } 2316 2317 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2318 { 2319 int mask; 2320 int ret; 2321 2322 addr += (nr >> 3); 2323 mask = 1 << (7 - (nr & 0x07)); 2324 ret = mask & *addr; 2325 *addr |= mask; 2326 return ret; 2327 } 2328 2329 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2330 { 2331 int mask; 2332 int ret; 2333 2334 addr += (nr >> 3); 2335 mask = 1 << (7 - (nr & 0x07)); 2336 ret = mask & *addr; 2337 *addr &= ~mask; 2338 return ret; 2339 } 2340 2341 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2342 { 2343 int mask; 2344 2345 addr += (nr >> 3); 2346 mask = 1 << (7 - (nr & 0x07)); 2347 *addr ^= mask; 2348 } 2349 2350 /* 2351 * On-disk inode flags (f2fs_inode::i_flags) 2352 */ 2353 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2354 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2355 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2356 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2357 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2358 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2359 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2360 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2361 2362 /* Flags that should be inherited by new inodes from their parent. */ 2363 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2364 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL) 2365 2366 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2367 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL)) 2368 2369 /* Flags that are appropriate for non-directories/regular files. */ 2370 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2371 2372 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2373 { 2374 if (S_ISDIR(mode)) 2375 return flags; 2376 else if (S_ISREG(mode)) 2377 return flags & F2FS_REG_FLMASK; 2378 else 2379 return flags & F2FS_OTHER_FLMASK; 2380 } 2381 2382 /* used for f2fs_inode_info->flags */ 2383 enum { 2384 FI_NEW_INODE, /* indicate newly allocated inode */ 2385 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 2386 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 2387 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 2388 FI_INC_LINK, /* need to increment i_nlink */ 2389 FI_ACL_MODE, /* indicate acl mode */ 2390 FI_NO_ALLOC, /* should not allocate any blocks */ 2391 FI_FREE_NID, /* free allocated nide */ 2392 FI_NO_EXTENT, /* not to use the extent cache */ 2393 FI_INLINE_XATTR, /* used for inline xattr */ 2394 FI_INLINE_DATA, /* used for inline data*/ 2395 FI_INLINE_DENTRY, /* used for inline dentry */ 2396 FI_APPEND_WRITE, /* inode has appended data */ 2397 FI_UPDATE_WRITE, /* inode has in-place-update data */ 2398 FI_NEED_IPU, /* used for ipu per file */ 2399 FI_ATOMIC_FILE, /* indicate atomic file */ 2400 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 2401 FI_VOLATILE_FILE, /* indicate volatile file */ 2402 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 2403 FI_DROP_CACHE, /* drop dirty page cache */ 2404 FI_DATA_EXIST, /* indicate data exists */ 2405 FI_INLINE_DOTS, /* indicate inline dot dentries */ 2406 FI_DO_DEFRAG, /* indicate defragment is running */ 2407 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 2408 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 2409 FI_HOT_DATA, /* indicate file is hot */ 2410 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 2411 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 2412 FI_PIN_FILE, /* indicate file should not be gced */ 2413 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 2414 }; 2415 2416 static inline void __mark_inode_dirty_flag(struct inode *inode, 2417 int flag, bool set) 2418 { 2419 switch (flag) { 2420 case FI_INLINE_XATTR: 2421 case FI_INLINE_DATA: 2422 case FI_INLINE_DENTRY: 2423 case FI_NEW_INODE: 2424 if (set) 2425 return; 2426 /* fall through */ 2427 case FI_DATA_EXIST: 2428 case FI_INLINE_DOTS: 2429 case FI_PIN_FILE: 2430 f2fs_mark_inode_dirty_sync(inode, true); 2431 } 2432 } 2433 2434 static inline void set_inode_flag(struct inode *inode, int flag) 2435 { 2436 if (!test_bit(flag, &F2FS_I(inode)->flags)) 2437 set_bit(flag, &F2FS_I(inode)->flags); 2438 __mark_inode_dirty_flag(inode, flag, true); 2439 } 2440 2441 static inline int is_inode_flag_set(struct inode *inode, int flag) 2442 { 2443 return test_bit(flag, &F2FS_I(inode)->flags); 2444 } 2445 2446 static inline void clear_inode_flag(struct inode *inode, int flag) 2447 { 2448 if (test_bit(flag, &F2FS_I(inode)->flags)) 2449 clear_bit(flag, &F2FS_I(inode)->flags); 2450 __mark_inode_dirty_flag(inode, flag, false); 2451 } 2452 2453 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2454 { 2455 F2FS_I(inode)->i_acl_mode = mode; 2456 set_inode_flag(inode, FI_ACL_MODE); 2457 f2fs_mark_inode_dirty_sync(inode, false); 2458 } 2459 2460 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2461 { 2462 if (inc) 2463 inc_nlink(inode); 2464 else 2465 drop_nlink(inode); 2466 f2fs_mark_inode_dirty_sync(inode, true); 2467 } 2468 2469 static inline void f2fs_i_blocks_write(struct inode *inode, 2470 block_t diff, bool add, bool claim) 2471 { 2472 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2473 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2474 2475 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2476 if (add) { 2477 if (claim) 2478 dquot_claim_block(inode, diff); 2479 else 2480 dquot_alloc_block_nofail(inode, diff); 2481 } else { 2482 dquot_free_block(inode, diff); 2483 } 2484 2485 f2fs_mark_inode_dirty_sync(inode, true); 2486 if (clean || recover) 2487 set_inode_flag(inode, FI_AUTO_RECOVER); 2488 } 2489 2490 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2491 { 2492 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2493 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2494 2495 if (i_size_read(inode) == i_size) 2496 return; 2497 2498 i_size_write(inode, i_size); 2499 f2fs_mark_inode_dirty_sync(inode, true); 2500 if (clean || recover) 2501 set_inode_flag(inode, FI_AUTO_RECOVER); 2502 } 2503 2504 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2505 { 2506 F2FS_I(inode)->i_current_depth = depth; 2507 f2fs_mark_inode_dirty_sync(inode, true); 2508 } 2509 2510 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2511 unsigned int count) 2512 { 2513 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2514 f2fs_mark_inode_dirty_sync(inode, true); 2515 } 2516 2517 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2518 { 2519 F2FS_I(inode)->i_xattr_nid = xnid; 2520 f2fs_mark_inode_dirty_sync(inode, true); 2521 } 2522 2523 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2524 { 2525 F2FS_I(inode)->i_pino = pino; 2526 f2fs_mark_inode_dirty_sync(inode, true); 2527 } 2528 2529 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2530 { 2531 struct f2fs_inode_info *fi = F2FS_I(inode); 2532 2533 if (ri->i_inline & F2FS_INLINE_XATTR) 2534 set_bit(FI_INLINE_XATTR, &fi->flags); 2535 if (ri->i_inline & F2FS_INLINE_DATA) 2536 set_bit(FI_INLINE_DATA, &fi->flags); 2537 if (ri->i_inline & F2FS_INLINE_DENTRY) 2538 set_bit(FI_INLINE_DENTRY, &fi->flags); 2539 if (ri->i_inline & F2FS_DATA_EXIST) 2540 set_bit(FI_DATA_EXIST, &fi->flags); 2541 if (ri->i_inline & F2FS_INLINE_DOTS) 2542 set_bit(FI_INLINE_DOTS, &fi->flags); 2543 if (ri->i_inline & F2FS_EXTRA_ATTR) 2544 set_bit(FI_EXTRA_ATTR, &fi->flags); 2545 if (ri->i_inline & F2FS_PIN_FILE) 2546 set_bit(FI_PIN_FILE, &fi->flags); 2547 } 2548 2549 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2550 { 2551 ri->i_inline = 0; 2552 2553 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2554 ri->i_inline |= F2FS_INLINE_XATTR; 2555 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2556 ri->i_inline |= F2FS_INLINE_DATA; 2557 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2558 ri->i_inline |= F2FS_INLINE_DENTRY; 2559 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2560 ri->i_inline |= F2FS_DATA_EXIST; 2561 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2562 ri->i_inline |= F2FS_INLINE_DOTS; 2563 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2564 ri->i_inline |= F2FS_EXTRA_ATTR; 2565 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2566 ri->i_inline |= F2FS_PIN_FILE; 2567 } 2568 2569 static inline int f2fs_has_extra_attr(struct inode *inode) 2570 { 2571 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2572 } 2573 2574 static inline int f2fs_has_inline_xattr(struct inode *inode) 2575 { 2576 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2577 } 2578 2579 static inline unsigned int addrs_per_inode(struct inode *inode) 2580 { 2581 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2582 get_inline_xattr_addrs(inode); 2583 return ALIGN_DOWN(addrs, 1); 2584 } 2585 2586 static inline unsigned int addrs_per_block(struct inode *inode) 2587 { 2588 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, 1); 2589 } 2590 2591 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2592 { 2593 struct f2fs_inode *ri = F2FS_INODE(page); 2594 2595 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2596 get_inline_xattr_addrs(inode)]); 2597 } 2598 2599 static inline int inline_xattr_size(struct inode *inode) 2600 { 2601 if (f2fs_has_inline_xattr(inode)) 2602 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2603 return 0; 2604 } 2605 2606 static inline int f2fs_has_inline_data(struct inode *inode) 2607 { 2608 return is_inode_flag_set(inode, FI_INLINE_DATA); 2609 } 2610 2611 static inline int f2fs_exist_data(struct inode *inode) 2612 { 2613 return is_inode_flag_set(inode, FI_DATA_EXIST); 2614 } 2615 2616 static inline int f2fs_has_inline_dots(struct inode *inode) 2617 { 2618 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2619 } 2620 2621 static inline bool f2fs_is_pinned_file(struct inode *inode) 2622 { 2623 return is_inode_flag_set(inode, FI_PIN_FILE); 2624 } 2625 2626 static inline bool f2fs_is_atomic_file(struct inode *inode) 2627 { 2628 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2629 } 2630 2631 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2632 { 2633 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2634 } 2635 2636 static inline bool f2fs_is_volatile_file(struct inode *inode) 2637 { 2638 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2639 } 2640 2641 static inline bool f2fs_is_first_block_written(struct inode *inode) 2642 { 2643 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2644 } 2645 2646 static inline bool f2fs_is_drop_cache(struct inode *inode) 2647 { 2648 return is_inode_flag_set(inode, FI_DROP_CACHE); 2649 } 2650 2651 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2652 { 2653 struct f2fs_inode *ri = F2FS_INODE(page); 2654 int extra_size = get_extra_isize(inode); 2655 2656 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2657 } 2658 2659 static inline int f2fs_has_inline_dentry(struct inode *inode) 2660 { 2661 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2662 } 2663 2664 static inline int is_file(struct inode *inode, int type) 2665 { 2666 return F2FS_I(inode)->i_advise & type; 2667 } 2668 2669 static inline void set_file(struct inode *inode, int type) 2670 { 2671 F2FS_I(inode)->i_advise |= type; 2672 f2fs_mark_inode_dirty_sync(inode, true); 2673 } 2674 2675 static inline void clear_file(struct inode *inode, int type) 2676 { 2677 F2FS_I(inode)->i_advise &= ~type; 2678 f2fs_mark_inode_dirty_sync(inode, true); 2679 } 2680 2681 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2682 { 2683 bool ret; 2684 2685 if (dsync) { 2686 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2687 2688 spin_lock(&sbi->inode_lock[DIRTY_META]); 2689 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2690 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2691 return ret; 2692 } 2693 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2694 file_keep_isize(inode) || 2695 i_size_read(inode) & ~PAGE_MASK) 2696 return false; 2697 2698 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2699 return false; 2700 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2701 return false; 2702 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2703 return false; 2704 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2705 &F2FS_I(inode)->i_crtime)) 2706 return false; 2707 2708 down_read(&F2FS_I(inode)->i_sem); 2709 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2710 up_read(&F2FS_I(inode)->i_sem); 2711 2712 return ret; 2713 } 2714 2715 static inline bool f2fs_readonly(struct super_block *sb) 2716 { 2717 return sb_rdonly(sb); 2718 } 2719 2720 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2721 { 2722 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2723 } 2724 2725 static inline bool is_dot_dotdot(const struct qstr *str) 2726 { 2727 if (str->len == 1 && str->name[0] == '.') 2728 return true; 2729 2730 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.') 2731 return true; 2732 2733 return false; 2734 } 2735 2736 static inline bool f2fs_may_extent_tree(struct inode *inode) 2737 { 2738 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2739 2740 if (!test_opt(sbi, EXTENT_CACHE) || 2741 is_inode_flag_set(inode, FI_NO_EXTENT)) 2742 return false; 2743 2744 /* 2745 * for recovered files during mount do not create extents 2746 * if shrinker is not registered. 2747 */ 2748 if (list_empty(&sbi->s_list)) 2749 return false; 2750 2751 return S_ISREG(inode->i_mode); 2752 } 2753 2754 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2755 size_t size, gfp_t flags) 2756 { 2757 void *ret; 2758 2759 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2760 f2fs_show_injection_info(FAULT_KMALLOC); 2761 return NULL; 2762 } 2763 2764 ret = kmalloc(size, flags); 2765 if (ret) 2766 return ret; 2767 2768 return kvmalloc(size, flags); 2769 } 2770 2771 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2772 size_t size, gfp_t flags) 2773 { 2774 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2775 } 2776 2777 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2778 size_t size, gfp_t flags) 2779 { 2780 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2781 f2fs_show_injection_info(FAULT_KVMALLOC); 2782 return NULL; 2783 } 2784 2785 return kvmalloc(size, flags); 2786 } 2787 2788 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2789 size_t size, gfp_t flags) 2790 { 2791 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2792 } 2793 2794 static inline int get_extra_isize(struct inode *inode) 2795 { 2796 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2797 } 2798 2799 static inline int get_inline_xattr_addrs(struct inode *inode) 2800 { 2801 return F2FS_I(inode)->i_inline_xattr_size; 2802 } 2803 2804 #define f2fs_get_inode_mode(i) \ 2805 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2806 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2807 2808 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2809 (offsetof(struct f2fs_inode, i_extra_end) - \ 2810 offsetof(struct f2fs_inode, i_extra_isize)) \ 2811 2812 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2813 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 2814 ((offsetof(typeof(*(f2fs_inode)), field) + \ 2815 sizeof((f2fs_inode)->field)) \ 2816 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 2817 2818 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 2819 { 2820 int i; 2821 2822 spin_lock(&sbi->iostat_lock); 2823 for (i = 0; i < NR_IO_TYPE; i++) 2824 sbi->write_iostat[i] = 0; 2825 spin_unlock(&sbi->iostat_lock); 2826 } 2827 2828 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 2829 enum iostat_type type, unsigned long long io_bytes) 2830 { 2831 if (!sbi->iostat_enable) 2832 return; 2833 spin_lock(&sbi->iostat_lock); 2834 sbi->write_iostat[type] += io_bytes; 2835 2836 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 2837 sbi->write_iostat[APP_BUFFERED_IO] = 2838 sbi->write_iostat[APP_WRITE_IO] - 2839 sbi->write_iostat[APP_DIRECT_IO]; 2840 spin_unlock(&sbi->iostat_lock); 2841 } 2842 2843 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 2844 2845 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 2846 2847 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 2848 block_t blkaddr, int type); 2849 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 2850 block_t blkaddr, int type) 2851 { 2852 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 2853 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 2854 blkaddr, type); 2855 f2fs_bug_on(sbi, 1); 2856 } 2857 } 2858 2859 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 2860 { 2861 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR) 2862 return false; 2863 return true; 2864 } 2865 2866 static inline void f2fs_set_page_private(struct page *page, 2867 unsigned long data) 2868 { 2869 if (PagePrivate(page)) 2870 return; 2871 2872 get_page(page); 2873 SetPagePrivate(page); 2874 set_page_private(page, data); 2875 } 2876 2877 static inline void f2fs_clear_page_private(struct page *page) 2878 { 2879 if (!PagePrivate(page)) 2880 return; 2881 2882 set_page_private(page, 0); 2883 ClearPagePrivate(page); 2884 f2fs_put_page(page, 0); 2885 } 2886 2887 /* 2888 * file.c 2889 */ 2890 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 2891 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 2892 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 2893 int f2fs_truncate(struct inode *inode); 2894 int f2fs_getattr(const struct path *path, struct kstat *stat, 2895 u32 request_mask, unsigned int flags); 2896 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 2897 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 2898 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 2899 int f2fs_precache_extents(struct inode *inode); 2900 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 2901 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 2902 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 2903 int f2fs_pin_file_control(struct inode *inode, bool inc); 2904 2905 /* 2906 * inode.c 2907 */ 2908 void f2fs_set_inode_flags(struct inode *inode); 2909 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 2910 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 2911 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 2912 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 2913 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 2914 void f2fs_update_inode(struct inode *inode, struct page *node_page); 2915 void f2fs_update_inode_page(struct inode *inode); 2916 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 2917 void f2fs_evict_inode(struct inode *inode); 2918 void f2fs_handle_failed_inode(struct inode *inode); 2919 2920 /* 2921 * namei.c 2922 */ 2923 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 2924 bool hot, bool set); 2925 struct dentry *f2fs_get_parent(struct dentry *child); 2926 2927 /* 2928 * dir.c 2929 */ 2930 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 2931 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname, 2932 f2fs_hash_t namehash, int *max_slots, 2933 struct f2fs_dentry_ptr *d); 2934 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 2935 unsigned int start_pos, struct fscrypt_str *fstr); 2936 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 2937 struct f2fs_dentry_ptr *d); 2938 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 2939 const struct qstr *new_name, 2940 const struct qstr *orig_name, struct page *dpage); 2941 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 2942 unsigned int current_depth); 2943 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 2944 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 2945 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 2946 struct fscrypt_name *fname, struct page **res_page); 2947 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 2948 const struct qstr *child, struct page **res_page); 2949 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 2950 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 2951 struct page **page); 2952 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 2953 struct page *page, struct inode *inode); 2954 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 2955 const struct qstr *name, f2fs_hash_t name_hash, 2956 unsigned int bit_pos); 2957 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, 2958 const struct qstr *orig_name, 2959 struct inode *inode, nid_t ino, umode_t mode); 2960 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname, 2961 struct inode *inode, nid_t ino, umode_t mode); 2962 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 2963 struct inode *inode, nid_t ino, umode_t mode); 2964 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 2965 struct inode *dir, struct inode *inode); 2966 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 2967 bool f2fs_empty_dir(struct inode *dir); 2968 2969 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 2970 { 2971 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 2972 inode, inode->i_ino, inode->i_mode); 2973 } 2974 2975 /* 2976 * super.c 2977 */ 2978 int f2fs_inode_dirtied(struct inode *inode, bool sync); 2979 void f2fs_inode_synced(struct inode *inode); 2980 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 2981 int f2fs_quota_sync(struct super_block *sb, int type); 2982 void f2fs_quota_off_umount(struct super_block *sb); 2983 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 2984 int f2fs_sync_fs(struct super_block *sb, int sync); 2985 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 2986 2987 /* 2988 * hash.c 2989 */ 2990 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info, 2991 struct fscrypt_name *fname); 2992 2993 /* 2994 * node.c 2995 */ 2996 struct dnode_of_data; 2997 struct node_info; 2998 2999 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3000 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3001 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3002 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3003 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3004 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3005 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3006 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3007 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3008 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3009 struct node_info *ni); 3010 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3011 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3012 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3013 int f2fs_truncate_xattr_node(struct inode *inode); 3014 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3015 unsigned int seq_id); 3016 int f2fs_remove_inode_page(struct inode *inode); 3017 struct page *f2fs_new_inode_page(struct inode *inode); 3018 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3019 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3020 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3021 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3022 int f2fs_move_node_page(struct page *node_page, int gc_type); 3023 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3024 struct writeback_control *wbc, bool atomic, 3025 unsigned int *seq_id); 3026 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3027 struct writeback_control *wbc, 3028 bool do_balance, enum iostat_type io_type); 3029 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3030 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3031 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3032 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3033 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3034 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3035 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3036 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3037 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3038 unsigned int segno, struct f2fs_summary_block *sum); 3039 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3040 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3041 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3042 int __init f2fs_create_node_manager_caches(void); 3043 void f2fs_destroy_node_manager_caches(void); 3044 3045 /* 3046 * segment.c 3047 */ 3048 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3049 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3050 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3051 void f2fs_drop_inmem_pages(struct inode *inode); 3052 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3053 int f2fs_commit_inmem_pages(struct inode *inode); 3054 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3055 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi); 3056 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3057 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3058 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3059 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3060 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3061 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3062 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3063 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3064 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3065 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3066 struct cp_control *cpc); 3067 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3068 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3069 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3070 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3071 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3072 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3073 unsigned int start, unsigned int end); 3074 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3075 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3076 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3077 struct cp_control *cpc); 3078 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3079 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3080 block_t blk_addr); 3081 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3082 enum iostat_type io_type); 3083 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3084 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3085 struct f2fs_io_info *fio); 3086 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3087 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3088 block_t old_blkaddr, block_t new_blkaddr, 3089 bool recover_curseg, bool recover_newaddr); 3090 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3091 block_t old_addr, block_t new_addr, 3092 unsigned char version, bool recover_curseg, 3093 bool recover_newaddr); 3094 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3095 block_t old_blkaddr, block_t *new_blkaddr, 3096 struct f2fs_summary *sum, int type, 3097 struct f2fs_io_info *fio, bool add_list); 3098 void f2fs_wait_on_page_writeback(struct page *page, 3099 enum page_type type, bool ordered, bool locked); 3100 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3101 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3102 block_t len); 3103 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3104 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3105 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3106 unsigned int val, int alloc); 3107 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3108 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3109 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3110 int __init f2fs_create_segment_manager_caches(void); 3111 void f2fs_destroy_segment_manager_caches(void); 3112 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3113 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3114 enum page_type type, enum temp_type temp); 3115 3116 /* 3117 * checkpoint.c 3118 */ 3119 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3120 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3121 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3122 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3123 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3124 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3125 block_t blkaddr, int type); 3126 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3127 int type, bool sync); 3128 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3129 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3130 long nr_to_write, enum iostat_type io_type); 3131 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3132 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3133 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3134 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3135 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3136 unsigned int devidx, int type); 3137 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3138 unsigned int devidx, int type); 3139 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3140 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3141 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3142 void f2fs_add_orphan_inode(struct inode *inode); 3143 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3144 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3145 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3146 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3147 void f2fs_remove_dirty_inode(struct inode *inode); 3148 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3149 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi); 3150 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3151 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3152 int __init f2fs_create_checkpoint_caches(void); 3153 void f2fs_destroy_checkpoint_caches(void); 3154 3155 /* 3156 * data.c 3157 */ 3158 int f2fs_init_post_read_processing(void); 3159 void f2fs_destroy_post_read_processing(void); 3160 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3161 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3162 struct inode *inode, struct page *page, 3163 nid_t ino, enum page_type type); 3164 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3165 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3166 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3167 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3168 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3169 block_t blk_addr, struct bio *bio); 3170 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3171 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3172 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3173 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3174 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3175 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3176 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3177 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3178 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3179 int op_flags, bool for_write); 3180 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3181 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3182 bool for_write); 3183 struct page *f2fs_get_new_data_page(struct inode *inode, 3184 struct page *ipage, pgoff_t index, bool new_i_size); 3185 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3186 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3187 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3188 int create, int flag); 3189 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3190 u64 start, u64 len); 3191 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3192 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3193 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3194 unsigned int length); 3195 int f2fs_release_page(struct page *page, gfp_t wait); 3196 #ifdef CONFIG_MIGRATION 3197 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3198 struct page *page, enum migrate_mode mode); 3199 #endif 3200 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3201 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3202 3203 /* 3204 * gc.c 3205 */ 3206 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3207 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3208 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3209 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3210 unsigned int segno); 3211 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3212 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3213 3214 /* 3215 * recovery.c 3216 */ 3217 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3218 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3219 3220 /* 3221 * debug.c 3222 */ 3223 #ifdef CONFIG_F2FS_STAT_FS 3224 struct f2fs_stat_info { 3225 struct list_head stat_list; 3226 struct f2fs_sb_info *sbi; 3227 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3228 int main_area_segs, main_area_sections, main_area_zones; 3229 unsigned long long hit_largest, hit_cached, hit_rbtree; 3230 unsigned long long hit_total, total_ext; 3231 int ext_tree, zombie_tree, ext_node; 3232 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3233 int ndirty_data, ndirty_qdata; 3234 int inmem_pages; 3235 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3236 int nats, dirty_nats, sits, dirty_sits; 3237 int free_nids, avail_nids, alloc_nids; 3238 int total_count, utilization; 3239 int bg_gc, nr_wb_cp_data, nr_wb_data; 3240 int nr_rd_data, nr_rd_node, nr_rd_meta; 3241 int nr_dio_read, nr_dio_write; 3242 unsigned int io_skip_bggc, other_skip_bggc; 3243 int nr_flushing, nr_flushed, flush_list_empty; 3244 int nr_discarding, nr_discarded; 3245 int nr_discard_cmd; 3246 unsigned int undiscard_blks; 3247 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3248 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3249 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3250 unsigned int bimodal, avg_vblocks; 3251 int util_free, util_valid, util_invalid; 3252 int rsvd_segs, overp_segs; 3253 int dirty_count, node_pages, meta_pages; 3254 int prefree_count, call_count, cp_count, bg_cp_count; 3255 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3256 int bg_node_segs, bg_data_segs; 3257 int tot_blks, data_blks, node_blks; 3258 int bg_data_blks, bg_node_blks; 3259 unsigned long long skipped_atomic_files[2]; 3260 int curseg[NR_CURSEG_TYPE]; 3261 int cursec[NR_CURSEG_TYPE]; 3262 int curzone[NR_CURSEG_TYPE]; 3263 3264 unsigned int meta_count[META_MAX]; 3265 unsigned int segment_count[2]; 3266 unsigned int block_count[2]; 3267 unsigned int inplace_count; 3268 unsigned long long base_mem, cache_mem, page_mem; 3269 }; 3270 3271 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3272 { 3273 return (struct f2fs_stat_info *)sbi->stat_info; 3274 } 3275 3276 #define stat_inc_cp_count(si) ((si)->cp_count++) 3277 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3278 #define stat_inc_call_count(si) ((si)->call_count++) 3279 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++) 3280 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3281 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3282 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3283 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3284 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3285 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3286 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3287 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3288 #define stat_inc_inline_xattr(inode) \ 3289 do { \ 3290 if (f2fs_has_inline_xattr(inode)) \ 3291 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3292 } while (0) 3293 #define stat_dec_inline_xattr(inode) \ 3294 do { \ 3295 if (f2fs_has_inline_xattr(inode)) \ 3296 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3297 } while (0) 3298 #define stat_inc_inline_inode(inode) \ 3299 do { \ 3300 if (f2fs_has_inline_data(inode)) \ 3301 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3302 } while (0) 3303 #define stat_dec_inline_inode(inode) \ 3304 do { \ 3305 if (f2fs_has_inline_data(inode)) \ 3306 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3307 } while (0) 3308 #define stat_inc_inline_dir(inode) \ 3309 do { \ 3310 if (f2fs_has_inline_dentry(inode)) \ 3311 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3312 } while (0) 3313 #define stat_dec_inline_dir(inode) \ 3314 do { \ 3315 if (f2fs_has_inline_dentry(inode)) \ 3316 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3317 } while (0) 3318 #define stat_inc_meta_count(sbi, blkaddr) \ 3319 do { \ 3320 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3321 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3322 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3323 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3324 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3325 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3326 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3327 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3328 } while (0) 3329 #define stat_inc_seg_type(sbi, curseg) \ 3330 ((sbi)->segment_count[(curseg)->alloc_type]++) 3331 #define stat_inc_block_count(sbi, curseg) \ 3332 ((sbi)->block_count[(curseg)->alloc_type]++) 3333 #define stat_inc_inplace_blocks(sbi) \ 3334 (atomic_inc(&(sbi)->inplace_count)) 3335 #define stat_inc_atomic_write(inode) \ 3336 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt)) 3337 #define stat_dec_atomic_write(inode) \ 3338 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt)) 3339 #define stat_update_max_atomic_write(inode) \ 3340 do { \ 3341 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \ 3342 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3343 if (cur > max) \ 3344 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3345 } while (0) 3346 #define stat_inc_volatile_write(inode) \ 3347 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3348 #define stat_dec_volatile_write(inode) \ 3349 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3350 #define stat_update_max_volatile_write(inode) \ 3351 do { \ 3352 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3353 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3354 if (cur > max) \ 3355 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3356 } while (0) 3357 #define stat_inc_seg_count(sbi, type, gc_type) \ 3358 do { \ 3359 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3360 si->tot_segs++; \ 3361 if ((type) == SUM_TYPE_DATA) { \ 3362 si->data_segs++; \ 3363 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3364 } else { \ 3365 si->node_segs++; \ 3366 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3367 } \ 3368 } while (0) 3369 3370 #define stat_inc_tot_blk_count(si, blks) \ 3371 ((si)->tot_blks += (blks)) 3372 3373 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3374 do { \ 3375 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3376 stat_inc_tot_blk_count(si, blks); \ 3377 si->data_blks += (blks); \ 3378 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3379 } while (0) 3380 3381 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3382 do { \ 3383 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3384 stat_inc_tot_blk_count(si, blks); \ 3385 si->node_blks += (blks); \ 3386 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3387 } while (0) 3388 3389 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3390 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3391 void __init f2fs_create_root_stats(void); 3392 void f2fs_destroy_root_stats(void); 3393 #else 3394 #define stat_inc_cp_count(si) do { } while (0) 3395 #define stat_inc_bg_cp_count(si) do { } while (0) 3396 #define stat_inc_call_count(si) do { } while (0) 3397 #define stat_inc_bggc_count(si) do { } while (0) 3398 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3399 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3400 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3401 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3402 #define stat_inc_total_hit(sb) do { } while (0) 3403 #define stat_inc_rbtree_node_hit(sb) do { } while (0) 3404 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3405 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3406 #define stat_inc_inline_xattr(inode) do { } while (0) 3407 #define stat_dec_inline_xattr(inode) do { } while (0) 3408 #define stat_inc_inline_inode(inode) do { } while (0) 3409 #define stat_dec_inline_inode(inode) do { } while (0) 3410 #define stat_inc_inline_dir(inode) do { } while (0) 3411 #define stat_dec_inline_dir(inode) do { } while (0) 3412 #define stat_inc_atomic_write(inode) do { } while (0) 3413 #define stat_dec_atomic_write(inode) do { } while (0) 3414 #define stat_update_max_atomic_write(inode) do { } while (0) 3415 #define stat_inc_volatile_write(inode) do { } while (0) 3416 #define stat_dec_volatile_write(inode) do { } while (0) 3417 #define stat_update_max_volatile_write(inode) do { } while (0) 3418 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3419 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3420 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3421 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3422 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3423 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3424 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3425 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3426 3427 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3428 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3429 static inline void __init f2fs_create_root_stats(void) { } 3430 static inline void f2fs_destroy_root_stats(void) { } 3431 #endif 3432 3433 extern const struct file_operations f2fs_dir_operations; 3434 extern const struct file_operations f2fs_file_operations; 3435 extern const struct inode_operations f2fs_file_inode_operations; 3436 extern const struct address_space_operations f2fs_dblock_aops; 3437 extern const struct address_space_operations f2fs_node_aops; 3438 extern const struct address_space_operations f2fs_meta_aops; 3439 extern const struct inode_operations f2fs_dir_inode_operations; 3440 extern const struct inode_operations f2fs_symlink_inode_operations; 3441 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3442 extern const struct inode_operations f2fs_special_inode_operations; 3443 extern struct kmem_cache *f2fs_inode_entry_slab; 3444 3445 /* 3446 * inline.c 3447 */ 3448 bool f2fs_may_inline_data(struct inode *inode); 3449 bool f2fs_may_inline_dentry(struct inode *inode); 3450 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3451 void f2fs_truncate_inline_inode(struct inode *inode, 3452 struct page *ipage, u64 from); 3453 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3454 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3455 int f2fs_convert_inline_inode(struct inode *inode); 3456 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3457 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3458 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3459 struct fscrypt_name *fname, struct page **res_page); 3460 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3461 struct page *ipage); 3462 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name, 3463 const struct qstr *orig_name, 3464 struct inode *inode, nid_t ino, umode_t mode); 3465 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3466 struct page *page, struct inode *dir, 3467 struct inode *inode); 3468 bool f2fs_empty_inline_dir(struct inode *dir); 3469 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3470 struct fscrypt_str *fstr); 3471 int f2fs_inline_data_fiemap(struct inode *inode, 3472 struct fiemap_extent_info *fieinfo, 3473 __u64 start, __u64 len); 3474 3475 /* 3476 * shrinker.c 3477 */ 3478 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3479 struct shrink_control *sc); 3480 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3481 struct shrink_control *sc); 3482 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3483 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3484 3485 /* 3486 * extent_cache.c 3487 */ 3488 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3489 struct rb_entry *cached_re, unsigned int ofs); 3490 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3491 struct rb_root_cached *root, 3492 struct rb_node **parent, 3493 unsigned int ofs, bool *leftmost); 3494 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3495 struct rb_entry *cached_re, unsigned int ofs, 3496 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3497 struct rb_node ***insert_p, struct rb_node **insert_parent, 3498 bool force, bool *leftmost); 3499 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3500 struct rb_root_cached *root); 3501 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3502 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3503 void f2fs_drop_extent_tree(struct inode *inode); 3504 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3505 void f2fs_destroy_extent_tree(struct inode *inode); 3506 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3507 struct extent_info *ei); 3508 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3509 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3510 pgoff_t fofs, block_t blkaddr, unsigned int len); 3511 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3512 int __init f2fs_create_extent_cache(void); 3513 void f2fs_destroy_extent_cache(void); 3514 3515 /* 3516 * sysfs.c 3517 */ 3518 int __init f2fs_init_sysfs(void); 3519 void f2fs_exit_sysfs(void); 3520 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3521 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3522 3523 /* 3524 * crypto support 3525 */ 3526 static inline bool f2fs_encrypted_file(struct inode *inode) 3527 { 3528 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3529 } 3530 3531 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3532 { 3533 #ifdef CONFIG_FS_ENCRYPTION 3534 file_set_encrypt(inode); 3535 f2fs_set_inode_flags(inode); 3536 #endif 3537 } 3538 3539 /* 3540 * Returns true if the reads of the inode's data need to undergo some 3541 * postprocessing step, like decryption or authenticity verification. 3542 */ 3543 static inline bool f2fs_post_read_required(struct inode *inode) 3544 { 3545 return f2fs_encrypted_file(inode); 3546 } 3547 3548 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3549 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3550 { \ 3551 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3552 } 3553 3554 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3555 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3556 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3557 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3558 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3559 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3560 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3561 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3562 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3563 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3564 3565 #ifdef CONFIG_BLK_DEV_ZONED 3566 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3567 block_t blkaddr) 3568 { 3569 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3570 3571 return test_bit(zno, FDEV(devi).blkz_seq); 3572 } 3573 #endif 3574 3575 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3576 { 3577 return f2fs_sb_has_blkzoned(sbi); 3578 } 3579 3580 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3581 { 3582 return blk_queue_discard(bdev_get_queue(bdev)) || 3583 bdev_is_zoned(bdev); 3584 } 3585 3586 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3587 { 3588 int i; 3589 3590 if (!f2fs_is_multi_device(sbi)) 3591 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3592 3593 for (i = 0; i < sbi->s_ndevs; i++) 3594 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3595 return true; 3596 return false; 3597 } 3598 3599 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3600 { 3601 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3602 f2fs_hw_should_discard(sbi); 3603 } 3604 3605 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3606 { 3607 int i; 3608 3609 if (!f2fs_is_multi_device(sbi)) 3610 return bdev_read_only(sbi->sb->s_bdev); 3611 3612 for (i = 0; i < sbi->s_ndevs; i++) 3613 if (bdev_read_only(FDEV(i).bdev)) 3614 return true; 3615 return false; 3616 } 3617 3618 3619 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt) 3620 { 3621 clear_opt(sbi, ADAPTIVE); 3622 clear_opt(sbi, LFS); 3623 3624 switch (mt) { 3625 case F2FS_MOUNT_ADAPTIVE: 3626 set_opt(sbi, ADAPTIVE); 3627 break; 3628 case F2FS_MOUNT_LFS: 3629 set_opt(sbi, LFS); 3630 break; 3631 } 3632 } 3633 3634 static inline bool f2fs_may_encrypt(struct inode *inode) 3635 { 3636 #ifdef CONFIG_FS_ENCRYPTION 3637 umode_t mode = inode->i_mode; 3638 3639 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 3640 #else 3641 return false; 3642 #endif 3643 } 3644 3645 static inline int block_unaligned_IO(struct inode *inode, 3646 struct kiocb *iocb, struct iov_iter *iter) 3647 { 3648 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 3649 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 3650 loff_t offset = iocb->ki_pos; 3651 unsigned long align = offset | iov_iter_alignment(iter); 3652 3653 return align & blocksize_mask; 3654 } 3655 3656 static inline int allow_outplace_dio(struct inode *inode, 3657 struct kiocb *iocb, struct iov_iter *iter) 3658 { 3659 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3660 int rw = iov_iter_rw(iter); 3661 3662 return (test_opt(sbi, LFS) && (rw == WRITE) && 3663 !block_unaligned_IO(inode, iocb, iter)); 3664 } 3665 3666 static inline bool f2fs_force_buffered_io(struct inode *inode, 3667 struct kiocb *iocb, struct iov_iter *iter) 3668 { 3669 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3670 int rw = iov_iter_rw(iter); 3671 3672 if (f2fs_post_read_required(inode)) 3673 return true; 3674 if (f2fs_is_multi_device(sbi)) 3675 return true; 3676 /* 3677 * for blkzoned device, fallback direct IO to buffered IO, so 3678 * all IOs can be serialized by log-structured write. 3679 */ 3680 if (f2fs_sb_has_blkzoned(sbi)) 3681 return true; 3682 if (test_opt(sbi, LFS) && (rw == WRITE) && 3683 block_unaligned_IO(inode, iocb, iter)) 3684 return true; 3685 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 3686 return true; 3687 3688 return false; 3689 } 3690 3691 #ifdef CONFIG_F2FS_FAULT_INJECTION 3692 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 3693 unsigned int type); 3694 #else 3695 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 3696 #endif 3697 3698 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 3699 { 3700 #ifdef CONFIG_QUOTA 3701 if (f2fs_sb_has_quota_ino(sbi)) 3702 return true; 3703 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 3704 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 3705 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 3706 return true; 3707 #endif 3708 return false; 3709 } 3710 3711 #endif /* _LINUX_F2FS_H */ 3712